Generated power prediction system and method suitable for distributed photovoltaic power station

Abstract

The invention discloses a generated power prediction system and method suitable for a distributed photovoltaic power station, and relates to the technical field of generated power prediction. The system comprises a weather forecast server and a distributed photovoltaic power generation power prediction device, wherein the distributed photovoltaic power generation power prediction device comprises a data acquisition module, a data management module, a weather forecast data analysis module and a power generation power prediction module; power generation data and weather forecast data of a distributed photovoltaic power station are obtained from a photovoltaic local monitoring device or an SCADA system, weather forecast data from a weather forecast server and weather forecast data processed by a data management module are received and subjected to precision analysis, and weather forecast data with high precision are determined; and performing ultra-short-term, short-term and medium-term generation power prediction on the distributed photovoltaic power station based on the weather forecast data with high precision and the processed generation data output by the data management module. According to the invention, ultra-short-term, short-term and medium-term generation power prediction can be realized.

Description

Technical Field

[0001] This invention relates to the field of power generation prediction technology, and in particular to a power generation prediction system and method suitable for distributed photovoltaic power plants. Background Technology

[0002] By the first half of 2025, distributed photovoltaic (PV) capacity accounted for more than two-thirds of the total installed capacity, becoming a major source of electricity. Particularly in some regional power grids, over 70% of the power output during midday hours comes from distributed PV generation. This shift in power structure has transformed the power system from a traditional source-follow-load dynamic model to a dual-random load-source model, with the volatility of distributed PV output directly impacting the province's power balance. Power grid dispatching departments need to accurately grasp the distributed PV output curve 4 to 16 hours in advance; otherwise, they must reserve a large amount of spinning reserve capacity, leading to a significant increase in social energy costs.

[0003] Currently, output forecasting for distributed photovoltaic (PV) power plants mainly relies on basic monitoring systems and single meteorological data sources. Existing forecasting systems typically employ simple data processing methods and limited meteorological forecast sources, collecting power generation data through PV monitoring systems or SCADA systems and combining this data with a single meteorological forecast source for power prediction. These systems can achieve basic power generation data acquisition and short-term forecasting functions, and some systems also have data uploading capabilities, meeting the most basic power plant monitoring needs.

[0004] However, existing technologies have several shortcomings. The observability of distributed photovoltaic (PV) power plants is low; 10 kV distributed PV systems only upload 15 minutes of real-time power generation data and simple forecast data, lacking the ability to predict longer timescales. Meteorological conditions are highly uncertain; because data is obtained from only a single weather forecast source, the lack of a multi-source data comparison and analysis mechanism limits forecast accuracy. Communication quality is poor; limited by factors such as remote geographical locations, low compatibility of communication devices, aging sensors, and insufficient professional knowledge of maintenance personnel, the quality and completeness of power generation data are affected. Existing systems have weak data governance capabilities, lacking effective anomaly detection and handling mechanisms, and are unable to perform accuracy analysis and automatic optimization of multiple weather forecast sources, making it difficult to achieve high-precision multi-timescale power generation forecasting.

[0005] To address the aforementioned technical challenges, there is an urgent need to develop a novel distributed photovoltaic power generation prediction system that can effectively improve data quality, optimize the selection of meteorological forecast data sources, achieve accurate predictions across multiple time scales, and provide reliable technical support for power grid dispatching. Summary of the Invention

[0006] The technical problem to be solved by this invention is to address the shortcomings of existing technologies, specifically by providing a power generation prediction system and method suitable for distributed photovoltaic power plants, as detailed below: 1) In a first aspect, the present invention provides a power generation prediction system suitable for distributed photovoltaic power plants, comprising: a weather forecast server and a distributed photovoltaic power generation prediction device containing a data acquisition module, a data governance module, a weather forecast data analysis module, and a power generation prediction module; the weather forecast server is used to: acquire weather forecast data from multiple weather forecast websites; the data acquisition module is used to: acquire power generation data and weather forecast data of the distributed photovoltaic power plant from a local photovoltaic monitoring device or a SCADA system; the data governance module is used to: perform anomaly detection and processing on the power generation data and weather forecast data of the distributed photovoltaic power plant acquired by the data acquisition module; the weather forecast data analysis module is used to: receive and perform accuracy analysis on the weather forecast data from the weather forecast server and the processed weather forecast data from the data governance module, and determine the weather forecast data with high accuracy; the power generation prediction module is used to: predict the power generation of the distributed photovoltaic power plant in the ultra-short term, short term, and medium term based on the high-accuracy weather forecast data and the processed power generation data output by the data governance module.

[0007] The beneficial effects of the power generation prediction system for distributed photovoltaic power plants provided by this invention are as follows: The weather forecast server obtains weather forecast data from multiple weather forecast websites, overcoming the uncertainty of a single data source and providing rich input for prediction. The data acquisition module obtains power generation data and weather data from local photovoltaic monitoring devices or SCADA systems, ensuring comprehensive data sources. The data governance module performs anomaly detection and processing on the collected data, identifying and repairing outliers and noise, significantly improving data quality and resolving data integrity issues caused by communication equipment problems. The weather forecast data analysis module performs accuracy analysis on multi-source weather forecast data, automatically selecting high-accuracy weather forecast data to reduce the impact of weather condition uncertainty on prediction. Based on high-quality weather forecast data and the governed power generation data, the power generation prediction module achieves ultra-short-term, short-term, and medium-term power generation prediction, providing accurate output curves at multiple time scales. These improvements enable grid dispatch departments to anticipate changes in the output of distributed photovoltaic power plants, reduce spinning reserve capacity, lower social energy costs, and simultaneously improve power plant operating efficiency and grid compatibility.

[0008] 2) In a second aspect, the present invention also provides a method for predicting the power generation of a distributed photovoltaic power station, employing any one of the above-mentioned distributed photovoltaic power station power generation prediction systems. The method includes: acquiring weather forecast data from multiple weather forecast websites via a weather forecast server; acquiring power generation data and weather forecast data of the distributed photovoltaic power station from a local photovoltaic monitoring device or SCADA system via a data acquisition module; performing anomaly detection and processing on the power generation data and weather forecast data of the distributed photovoltaic power station acquired by the data acquisition module via a data governance module; receiving and performing accuracy analysis on the weather forecast data from the weather forecast server and the processed weather forecast data from the data governance module via a weather forecast data analysis module to determine the weather forecast data with high accuracy; and predicting the power generation of the distributed photovoltaic power station for ultra-short-term, short-term, and medium-term periods based on the high-accuracy weather forecast data and the processed power generation data output by the data governance module via a power generation prediction module.

[0009] 3) In a third aspect, the present invention also provides an electronic device, the electronic device including a processor coupled to a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor, so that the electronic device implements the above-mentioned method for predicting the power generation of a distributed photovoltaic power station.

[0010] 4) In a fourth aspect, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for predicting the power generation of a distributed photovoltaic power station.

[0011] It should be noted that the beneficial effects of the technical solutions of the second to fourth aspects of the present invention and their corresponding possible implementations can be found in the above description of the technical effects of the first aspect and its corresponding possible implementations, and will not be repeated here. Attached Figure Description

[0012] Figure 1 This is one of the structural schematic diagrams of a power generation prediction system suitable for distributed photovoltaic power stations according to an embodiment of the present invention; Figure 2 This is a second schematic diagram of a power generation prediction system for distributed photovoltaic power plants according to an embodiment of the present invention; Figure 3 A schematic diagram of a distributed photovoltaic power generation prediction device; Figure 4 This is a flowchart illustrating a power generation prediction method for distributed photovoltaic power plants according to an embodiment of the present invention. Detailed Implementation

[0013] like Figure 1 As shown, an embodiment of the present invention provides a power generation prediction system for distributed photovoltaic power plants, comprising: a weather forecast server and a distributed photovoltaic power generation prediction device containing a data acquisition module, a data governance module, a weather forecast data analysis module, and a power generation prediction module; Weather forecast servers are used to: retrieve weather forecast data from multiple weather forecast websites, specifically: 1) The weather forecast server pre-configures access information for multiple weather forecast websites, including Uniform Resource Locators (URIs), Application Programming Interface (API) keys, request headers, and timeout periods. The weather forecast server maintains a list of data sources, where each data source corresponds to a weather forecast website, and specifies the data update frequency, such as performing a data acquisition task every 30 minutes or hour. The weather forecast server uses a network library to initialize a secure Hypertext Transfer Protocol (HTTP) connection, ensuring encrypted data transmission.

[0014] 2) For each weather forecast website, the weather forecast server constructs a request message according to a predetermined schedule. The request message includes authentication parameters, such as an application programming interface key or token, and query parameters, such as geographic coordinates, time range, and weather element type. The weather forecast server sends the request to the weather forecast website using the Hypertext Transfer Protocol (HTTP) secure method and waits for a response. If a website does not respond, the weather forecast server logs the error and attempts to retry using an alternate Uniform Resource Locator (URL) or with a delay.

[0015] 3) Weather forecast websites return weather forecast data, typically transmitted in Extensible Markup Language (EXPLAIN), JavaScript object notation, or comma-separated value format. The weather forecast server checks the response status code, such as 200 for success, 404 for not found, and 500 for a server error. For a successful response, the weather forecast server reads the response body and verifies data integrity, such as checking for the existence of required fields. If the response fails, the weather forecast server triggers error handling procedures, such as using cached data or notifying the operations and maintenance system.

[0016] 4) The weather forecast server uses a parsing library to extract key fields from the weather forecast data, such as light intensity, temperature, humidity, wind speed, and wind direction. This data may be expressed in different units; for example, temperature may be provided in degrees Celsius or Fahrenheit. The weather forecast server performs unit conversion to ensure consistency. The parsed data is mapped to an internal standard format, including timestamps, data source identifiers, and meteorological element values. For example, light intensity is converted from watts per square meter to kilowatts per square meter using the following formula: ,in, This represents the standardized light intensity, expressed in kilowatts per square meter. This represents the original light intensity, measured in watts per square meter. This represents the conversion factor, with a value of 0.001.

[0017] 5) The parsed weather forecast data is stored in the server's memory or a local database, organized in time-series format. Each data point includes a timestamp, data source, and meteorological element values. The weather forecast server performs data quality checks, such as range validation, to ensure that light intensity is non-negative and temperature is within a reasonable range. If the data is invalid, the weather forecast server marks it as an anomaly and logs it.

[0018] 6) The weather forecast server converts the standardized weather forecast data into E-text file format, a plain text format where each line represents data for a single time point, with fields separated by delimiters. The weather forecast server generates a file header containing metadata such as file creation time, data source list, and version information. Then, the weather forecast server sends the E-text file to the reverse isolation device using a secure file transfer protocol through the firewall of the internet zone. During transmission, the weather forecast server verifies file integrity and size to ensure no data loss.

[0019] 7) The weather forecast server regularly updates data source configurations and handles website changes or application programming interface updates. The weather forecast server runs monitoring scripts to check the execution status of data acquisition tasks and network latency, and generates reports for system maintenance. If multiple acquisition failures occur consecutively, the server automatically disables the faulty data source and activates a backup plan.

[0020] The weather forecast server is a dedicated server device located in the Internet zone, responsible for acquiring weather forecast data from multiple weather forecast websites. The weather forecast server connects to the websites of various public or authorized weather data providers via the Internet, periodically collecting weather forecast information using network protocols and application programming interfaces (APIs). The weather forecast server has data request, reception, parsing, and temporary storage functions, can process various data formats, and converts raw data into a standard format usable by the system. The weather forecast server also communicates securely with the Internet zone's firewall and reverse isolation device to ensure that weather forecast data is not subject to unauthorized access or tampering during transmission.

[0021] Meteorological forecast data refers to predicted weather conditions obtained from meteorological forecast websites, including key meteorological elements such as sunlight intensity, temperature, humidity, wind speed, and wind direction. This data is represented digitally and typically includes time-series information, covering different time scales, such as short-term and medium-term forecasts. The accuracy and diversity of meteorological forecast data directly impact the power generation prediction accuracy of distributed photovoltaic power plants; therefore, the system needs to obtain data from multiple meteorological forecast websites to support subsequent comparative analysis and selection processes. Meteorological forecast data must maintain format consistency and integrity during transmission and parsing to ensure the reliability of the prediction model.

[0022] The data acquisition module is used to obtain power generation data and weather forecast data from distributed photovoltaic power plants from local photovoltaic monitoring devices or SCADA systems. The specific implementation process is as follows: 1) Upon startup, the data acquisition module loads a configuration file, defining the target data source information, including the network address, port number, communication protocol type, and authentication credentials of the photovoltaic local monitoring device or SCADA system. The data acquisition module supports multiple standard communication protocols, such as Modbus TCP, IEC-101 / 103 / 104, MQTT, DL / T645, and IEC-61850, and selects the appropriate protocol driver based on the configuration. The data acquisition module sets data request parameters, such as sampling frequency, data point list, and timeout, for example, requesting real-time power generation and solar irradiance data every 15 seconds. The data acquisition module also initializes an internal buffer and data queue for temporary storage of received data to prevent data loss.

[0023] 2) The data acquisition module establishes a network connection and a secure communication channel. It establishes a TCP / IP connection with the local photovoltaic monitoring device or SCADA system through the firewall in Security Zone I. The data acquisition module uses transport layer security protocols or virtual private network technology to encrypt the communication link, ensuring the confidentiality and integrity of data transmission. The data acquisition module verifies the identity of the target device through digital certificates or pre-shared keys. If the connection fails, the data acquisition module logs the error and retryes multiple times according to a retry policy, for example, retrying 3 times at 5-second intervals, until the connection is successful.

[0024] 3) The data acquisition module constructs and sends data request messages, generating standard request frames according to the selected communication protocol. For the Modbus TCP protocol, the data acquisition module constructs a request with function code 04 and reads the power generation value from the holding register. For the IEC-104 protocol, the data acquisition module sends a general call command, requesting all telemetry data. The request message includes the data point address, data length, and checksum information. The data acquisition module sends the request message to the photovoltaic local monitoring device or SCADA system through the established connection and starts a timer to monitor response timeouts.

[0025] 4) When the local photovoltaic monitoring device or SCADA system returns a response message, the data acquisition module reads the network stream and checks the message integrity. The data acquisition module uses a protocol parser to decode the response content and extract power generation data and weather forecast data from the distributed photovoltaic power station. For example, it parses power generation, voltage, current, irradiance, temperature, and humidity values ​​from the response frame. The data acquisition module verifies the data format and range; for example, power generation should be between zero and rated power. If the data is outside this range, it is marked as suspicious. The data acquisition module converts the parsed data into an internal standard format, including timestamps, data source identifiers, and data values.

[0026] 5) The data acquisition module performs preliminary quality control on the parsed data, such as checking timestamp continuity and data consistency. The data acquisition module calculates the data sampling rate using a formula to ensure it meets configuration requirements. The data acquisition module stores the data in a memory buffer or local database, awaiting further processing. The data acquisition module generates an event log to record the data acquisition status and any anomalies, such as communication interruptions or data errors. The data acquisition module forwards the data to the data governance module through an internal interface for further analysis.

[0027] 6) If a failure occurs during data acquisition, such as network interruption or device unresponsiveness, the data acquisition module triggers an error handling procedure. The data acquisition module attempts to switch to an alternative data source, such as switching from the SCADA system to the local photovoltaic monitoring device. The data acquisition module uses cached data to fill in missing values ​​to avoid predicting system outages. The data acquisition module periodically sends status reports to the system administrator, including data acquisition success rate and device health information.

[0028] The data acquisition module is a key component of the distributed photovoltaic (PV) power prediction device, specifically responsible for collecting relevant data from external systems. The data acquisition module communicates with devices in Safety Zone I via standard communication protocols, enabling real-time or periodic acquisition of power generation and meteorological data from the distributed PV power station. The module supports multiple protocols, including Modbus TCP, IEC-101 / 103 / 104, MQTT, DL / T645, and IEC-61850, ensuring compatibility with equipment from different manufacturers. The module also includes data buffering and error retry mechanisms to handle network latency or equipment failures, ensuring the continuity and reliability of data acquisition.

[0029] The photovoltaic (PV) local monitoring device is a monitoring equipment deployed in Safety Zone I of a distributed PV power station to monitor and control the operating status of the PV power generation units in real time. The PV local monitoring device collects power generation data from the distributed PV power station, such as voltage, current, and power output, as well as environmental meteorological data, such as light intensity, temperature, and humidity, through sensors and smart meters. The PV local monitoring device typically has local storage and display functions, capable of recording historical data and providing alarm information. The PV local monitoring device connects to the data acquisition module via a network, supporting standard industrial communication protocols to achieve data sharing and remote access.

[0030] SCADA system, short for Supervisory Control and Data Acquisition System, is a computer-controlled system used in industrial automation and widely applied in power systems. SCADA systems collect real-time operational data from distributed photovoltaic power plants, including power generation, equipment status, fault information, and meteorological monitoring data. SCADA systems provide a human-machine interface, allowing operators to perform remote control and parameter settings. The SCADA system and data acquisition modules communicate securely through a firewall, using standard protocols such as IEC-104 to transmit data, ensuring data integrity and real-time performance.

[0031] The power generation data of distributed photovoltaic (PV) power stations refers to the electrical parameters and performance indicators collected from PV power generation units, including real-time power generation, cumulative power generation, voltage, current, and frequency. This data is organized in time series format, covering different time scales, such as second-level, minute-level, or hourly data. The power generation data of distributed PV power stations also includes relevant meteorological information, such as solar irradiance, temperature, and wind speed, which are used to analyze power generation efficiency and predictive models. After being acquired by the data acquisition module, the power generation data of distributed PV power stations is used for subsequent data processing and predictive analysis, serving as the fundamental input for the power generation prediction system.

[0032] The data governance module is used to detect and process anomalies in the power generation data and weather forecast data of the distributed photovoltaic power station acquired by the data acquisition module. The specific implementation process is as follows: 1) The data governance module acquires power generation data and weather forecast data from the distributed photovoltaic power station from the data acquisition module. This data is transmitted in time-series format, including real-time and historical data. The data governance module initializes processing parameters, such as the sliding window size, statistical thresholds, and business rule base. The sliding window size is defined as the number of data points within a time interval; for example, using 30 consecutive data points as a window. The data governance module loads predefined business logic rules, including the rated installed capacity of the distributed photovoltaic power station, the maximum allowable rate of change, and the reasonable data range. The data governance module also sets up an internal buffer and log system for temporary data storage and recording of processing status.

[0033] 2) The data governance module performs anomaly detection based on statistical methods. Specifically, the data governance module calculates statistical indicators for each data point and uses a sliding window to analyze data distribution characteristics. For the power generation value in the power generation data, the data governance module calculates the mean of the data within the window. and standard deviation Mean This represents the mean and standard deviation of the data within the window. This indicates the degree of dispersion of the data. The data governance module uses the Z-score method to detect outliers, calculated as follows: ,in, Represents the Z-score value. This represents the value of the current data point. If... Greater than the preset threshold (For example If a data point is found to be out of the range, the data governance module marks it as an outlier. The same method is applied to light intensity values ​​in weather forecast data to ensure statistical consistency.

[0034] 3) The data governance module performs anomaly detection based on business logic. The data governance module applies business rules according to the operational characteristics of the distributed photovoltaic power station. For power generation data, the data governance module checks whether each power generation value exceeds the rated installed capacity. If the power generation value If a value is found to be outlier, it is marked as an outlier. The data governance module also calculates the rate of change between adjacent time points, using the following formula: ,in, Indicates the rate of change. Indicates a point in time The power generation value, This represents the power generation value at the previous time point. Indicates a time interval. If... Greater than the maximum allowable rate of change If the light intensity value is outside the reasonable range (e.g., 0 to 1500 W / m²) and the temperature value is outside the reasonable range (e.g., -50°C to 60°C), it is marked as an outlier. For weather forecast data, the data governance module checks whether the light intensity value is within the reasonable range (e.g., 0 to 1500 W / m²) and whether the temperature value is within the reasonable range (e.g., -50°C to 60°C). If it is outside the range, it is marked as an outlier.

[0035] 4) The data governance module further analyzes the marked outliers. It distinguishes between noise and genuine outliers, using a moving average filtering method to identify noise. The moving average calculation formula is: ,in, This represents the smoothed value. Indicates window size. Indicates the first in the window The data governance module compares the original values ​​with the smoothed values; if the deviation is less than the noise threshold... If the anomaly is positive, it is classified as noise; otherwise, it is classified as a genuine anomaly. The data governance module also analyzes the distribution pattern of outliers, such as continuous anomalies or isolated anomalies, to determine the handling strategy.

[0036] 5) The data governance module performs anomaly handling operations. Specifically, for data points confirmed as genuine anomalies, the data governance module takes different measures depending on the type. For data points that clearly do not conform to business logic, such as negative power generation values ​​or invalid values ​​in weather forecast data, the data governance module performs a deletion operation and removes them from the data sequence. For data points that can be replaced, the data governance module uses a linear interpolation method to calculate a reasonable value. The linear interpolation formula is: ,in, This represents the interpolated value. This indicates the previous normal value. This indicates the next normal value. Indicates the current time point, Indicates the previous point in time. This represents the next time point. If normal values ​​before and after are not available, the data governance module replaces them with the mean within the window, using the following formula: ,in, This represents the replacement value for the mean. This indicates the number of normal data points within the window. Indicates the first The values ​​of the normal data points.

[0037] 6) The data governance module verifies and outputs the processed data. The data governance module performs quality checks on the processed data to ensure that data points are within a reasonable range and that the time series is continuous. The data governance module generates a processing report, including the number of outliers, processing type, and data quality indicators. The processed distributed photovoltaic power station power generation data and weather forecast data are sent to subsequent modules, such as the power generation prediction module. The data governance module also updates its internal logs, recording detailed information on anomaly detection and handling for system monitoring and maintenance.

[0038] The weather forecast data analysis module is used to: receive and perform accuracy analysis on weather forecast data from the weather forecast server and processed weather forecast data from the data governance module, and determine the weather forecast data with high accuracy. Specifically: 1) The meteorological forecast data analysis module receives and integrates input data. It receives meteorological forecast data from multiple data sources, including predicted values ​​for key meteorological elements such as light intensity, temperature, humidity, wind speed, and wind direction, covering different time scales such as short-term and medium-term forecasts. Simultaneously, it receives processed real-time meteorological data and real-time power generation data from distributed photovoltaic power stations from the data governance module. This data has undergone anomaly detection and processing to ensure quality. The module initializes analysis parameters, including the time window size, error calculation method, and selected threshold. The time window size is defined as the analysis period; for example, data from the past 7 days may be used for accuracy assessment. The module aligns the meteorological forecast data from different data sources with the processed real-time data to ensure a one-to-one correspondence between data points, facilitating subsequent comparisons.

[0039] 2) The weather forecast data analysis module calculates the prediction error for each meteorological element. For each weather forecast data source, the module calculates the prediction error index for each meteorological element. The root mean square error (RMSE) is used as the primary error measure, and the formula is: ,in, This represents the root mean square error. Indicates the first Weather forecast data values ​​at each point in time, Indicates the first Actual observations at each time point (from real-time meteorological data processed by the data governance module). This represents the total number of data points within the time window. The weather forecast data analysis module also calculates the mean absolute error as an auxiliary indicator, using the following formula: ,in, This represents the mean absolute error. The weather forecast data analysis module calculates these error values ​​for each meteorological element (such as light intensity, temperature, humidity, wind speed, and wind direction) and stores the results for subsequent analysis.

[0040] 3) The weather forecast data analysis module performs correlation analysis between meteorological elements and power generation data. This module correlates the actual changes in various meteorological elements with the changes in real-time power generation data from distributed photovoltaic power plants to assess the explanatory power of weather forecast data on power generation variations. The Pearson correlation coefficient is used to measure the linear relationship; the formula is: ,in, Represents the correlation coefficient. Indicates the first Actual changes in meteorological elements (e.g., changes in light intensity) at a given time point. Indicates the first The change in power generation at each point in time, express The average value, express The average value, This indicates the number of data points. The weather forecast data analysis module calculates the correlation coefficient for each weather forecast data source and evaluates statistical significance, for example, by using a p-value test to ensure reliable correlation.

[0041] 4) The weather forecast data analysis module performs comprehensive accuracy scoring and ranking. Combining prediction errors and correlation analysis results, the module calculates a comprehensive accuracy score for each weather forecast data source. The comprehensive accuracy scoring formula is: ,in, This indicates the overall accuracy score. This represents the normalized root mean square error. Represents the correlation coefficient. and These are weighting coefficients, satisfying... The formula for calculating the normalized root mean square error is: ,in, This represents the minimum root mean square error among all weather forecast data sources. This represents the maximum root mean square error (RMSE) among all weather forecast data sources. The weather forecast data analysis module ranks the weather forecast data sources based on their comprehensive accuracy score and selects the data source with the highest score as the most accurate weather forecast data.

[0042] 5) The meteorological forecast data analysis module verifies and outputs selected data. The module performs a consistency check on the selected high-accuracy meteorological forecast data sources to ensure the data is continuous and without abrupt changes in the time series. The module outputs the selected results to the power generation prediction module and updates internal records, including accuracy indicators and the rationale for selection. The module periodically re-executes accuracy analysis, for example, every 24 hours, to adapt to changes in meteorological conditions and fluctuations in data source performance. The module generates an analysis report, including error values, correlation coefficients, and a comprehensive score, for system monitoring and optimization.

[0043] The power generation prediction module is used to predict the ultra-short-term, short-term, and medium-term power generation of distributed photovoltaic power stations based on high-precision meteorological forecast data and processed power generation data output from the data governance module. The specific implementation process is as follows: 1) The power generation prediction module prepares training and prediction data. It receives processed power generation data from the data governance module, including historical power generation values, voltage values, and current values, and high-precision weather forecast data from the weather forecast data analysis module, including light intensity, temperature, humidity, wind speed, and wind direction. The module performs feature engineering on this data, extracting time features such as hourly, daily, and monthly data, as well as meteorological features such as cumulative sunshine and average temperature. The module divides the data into training and testing sets; the training set is used for model learning, and the testing set is used for validation. Data standardization uses the Z-score method, with the formula: ,in, Represents the standardized eigenvalues. Represents the original feature values. Represents the characteristic mean. This represents the characteristic standard deviation. The power generation prediction module also handles missing values, using forward imputation or interpolation methods to ensure data integrity.

[0044] 2) Selection and Design of Deep Learning Model Architecture for Power Generation Prediction Module. The power generation prediction module adopts a Long Short-Term Memory (LSTM) network as its core model because LTM can effectively handle long-term dependencies in time series data. The LTM network unit structure includes an input gate, a forget gate, and an output gate. The calculation formula is as follows: in, Indicates the input gate output. Indicates the output of the forget gate. Indicates the output of the output gate. Indicates the state of candidate memory cells. Indicates the current state of memory cells. Indicates the current hidden state. Indicates the current input features. This indicates the previous hidden state. Represents the weight matrix. This represents the bias vector. This represents the sigmoid activation function. This represents the hyperbolic tangent activation function. The power generation prediction module adjusts the number of network layers and units according to the prediction time scale; for example, a shallower network is used for ultra-short-term predictions, while a deeper network is used for medium-term predictions. The power generation prediction module also integrates an attention mechanism to enhance focus on key time points.

[0045] 3) The power generation prediction module trains a deep learning model. The power generation prediction module uses training set data to train a Long Short-Term Memory (LSTM) network model to minimize prediction error. The loss function uses mean squared error, and the formula is: ,in, Indicates mean square error. This represents the actual power generation value. This represents the predicted power generation value from the model. This indicates the number of samples. The power generation prediction module uses an adaptive moment estimate optimizer for parameter updates, with a learning rate set to dynamically decay, for example, an initial value of 0.001. During training, the power generation prediction module applies an early stopping strategy to prevent overfitting; training stops when the validation set loss does not decrease for several consecutive cycles. The power generation prediction module saves the trained model parameters for subsequent prediction tasks.

[0046] 4) Power Generation Prediction Module Performance Verification. The power generation prediction module uses test set data to evaluate the trained model and calculate prediction accuracy metrics such as root mean square error (RMSE) and mean absolute error (MAE). The formula for RMSE is: ,in, This represents the root mean square error. Indicates the actual value. Indicates the predicted value. This indicates the number of test samples. The power generation prediction module compares the prediction errors at different time scales to ensure the model meets accuracy requirements. If performance is unsatisfactory, the power generation prediction module adjusts the model's hyperparameters, such as the hidden layer size or learning rate, and retrains the model.

[0047] 5) The power generation forecasting module performs multi-timescale power generation forecasting. The module loads a trained model and inputs real-time data, including high-precision weather forecasts and processed power generation data. For ultra-short-term forecasts (15 minutes to 4 hours), the module uses a rolling forecasting method, predicting 16 points (4 hours) in 15-minute intervals. For short-term forecasts (1 day to 3 days), the module predicts 72 points (3 days) in hourly intervals. For medium-term forecasts (4 days to 10 days), the module predicts 7 points (7 days, covering the 4-10 day range) in daily intervals. The forecast output is a power generation time series, with the formula: ,in, Indicates the future number Predicted power generation at each time point This represents a trained deep learning model. Represents the historical input feature sequence. This indicates the length of the input sequence. The power generation prediction module executes independently for each prediction time scale, ensuring that the output format meets grid requirements.

[0048] 6) Power Generation Forecasting Module Outputs and Stores Forecast Results. The power generation forecasting module sends ultra-short-term, short-term, and medium-term power generation forecasts to the application function modules for display and stores them in the database module. The power generation forecasting module periodically updates the model, such as retraining weekly, to adapt to changes in data distribution. The power generation forecasting module generates forecast reports, including confidence intervals and error analysis, for use in operation and maintenance decisions.

[0049] Ultra-short-term (USPS) power generation forecasting refers to predicting the power generation of distributed photovoltaic (PV) power plants within a timescale of 15 minutes to 4 hours. This short forecast timeframe, typically measured in minutes or hours, is used for real-time dispatching and rapid response to grid demands. USPS forecasting relies on high-frequency real-time data and recent weather forecasts to capture rapid fluctuations in power generation, such as those caused by changes in cloud cover. The output of USPS power generation forecasting is usually presented in time-series format, for example, with a forecast point every 15 minutes, helping maintenance personnel adjust operational strategies in a timely manner.

[0050] Short-term power generation forecasting refers to predicting the power generation of distributed photovoltaic (PV) power plants over a timescale of 1 to 3 days. This forecast covers diurnal variations, including daytime and nighttime power generation patterns, and is measured in hourly or daily units. Short-term forecasts are based on historical power generation data and medium- to long-term weather forecasts, taking into account the impact of seasonal and daily weather patterns. Short-term power generation forecasts are used for daily operational planning, energy trading, and grid load balancing, and the forecast results are typically output as daily or hourly power values.

[0051] Medium-term power generation forecasting refers to predicting the power generation of distributed photovoltaic (PV) power plants over a timescale of 4 to 10 days. This forecasting period is relatively long, focusing on weekly trends and measured in daily increments. Medium-term forecasts rely on extended weather forecast data and historical power generation patterns, enabling the identification of the impact of long-term weather systems such as high-pressure ridges or low-pressure troughs on power generation. Medium-term power generation forecasts are used to develop weekly power generation plans, maintenance arrangements, and resource allocation, with forecast results expressed as daily average or peak power values.

[0052] Optionally, in the above technical solution, the distributed photovoltaic power generation prediction device is located in Security Zone II, and the weather forecast server is located in the Internet Zone. The system also includes: a Beidou time synchronization device, a photovoltaic local monitoring device, and a Zone I station control switch located in Security Zone I; a first firewall connecting the Beidou time synchronization device, photovoltaic local monitoring device, and Zone I station control switch in Security Zone I with the equipment in Security Zone II; a Zone II switch located in Security Zone II; a reverse isolation device connecting the distributed photovoltaic power generation prediction device and Zone II switch in Security Zone II with the Internet Zone; and a second firewall located in the Internet Zone. The Beidou time synchronization device is used for Security Zone I... The equipment provides time synchronization services; the photovoltaic local monitoring device is used to monitor the operating status of the distributed photovoltaic power station and collect real-time power generation data; the Zone I station control switch connects the Beidou time synchronization device and the photovoltaic local monitoring device; the Zone II switch is used to connect the distributed photovoltaic power generation prediction device and the reverse isolation device; the weather forecast server is used to obtain weather forecast data from multiple weather forecast websites through the second firewall and transmit the weather forecast data to the reverse isolation device; the weather forecast data analysis module is specifically used to: receive and perform accuracy analysis on the weather forecast data from the reverse isolation device and the processed weather forecast data from the data governance module, and determine the weather forecast data with high accuracy.

[0053] To address the issues of poor communication quality, high data security risks, and limited meteorological data sources in distributed photovoltaic (PV) power plants, a rational system partitioning and equipment layout effectively improves overall operational efficiency. The distributed PV power generation forecasting device is located in Security Zone II, while the meteorological forecast server is located in the Internet Zone. Components such as a BeiDou time synchronization device, firewall, and reverse isolation device are configured to ensure secure isolation and reliable data transmission between zones. The BeiDou time synchronization device provides accurate time synchronization for devices in Security Zone I, ensuring data timestamp consistency and resolving data corruption caused by clock asynchrony. The first firewall and reverse isolation device enable controlled data exchange between Security Zone I and Security Zone II, and between Security Zone II and the Internet Zone, preventing external threats from intruding into the production control area and enhancing the system's anti-interference capabilities. The meteorological forecast server obtains data from multiple meteorological forecast websites via a second firewall and transmits the meteorological forecast data to the forecasting device through the reverse isolation device, supporting multi-source meteorological data acquisition and accuracy analysis. This architecture enhances data integrity and forecast reliability, providing high-quality input for power generation forecasting while reducing operational risks caused by communication interruptions or security vulnerabilities.

[0054] Optionally, in the above technical solution, the distributed photovoltaic power generation prediction device further includes a prediction data forwarding module. This module is used to: convert the ultra-short-term, short-term, and medium-term power generation prediction data generated by the power generation prediction module into the data format required by the power grid dispatch center, and upload it to the prediction master station through a dedicated power channel. Specifically: 1) The forecast data forwarding module receives and verifies power generation forecast data. It obtains ultra-short-term, short-term, and medium-term power generation forecast data from the power generation forecast module. This data is organized in time-series format and includes timestamps and predicted power values. The forecast data forwarding module performs integrity checks on the received data, verifying that the data format conforms to internal specifications, including checking timestamp continuity, data value validity, and sequence integrity. The forecast data forwarding module uses a data verification algorithm to ensure that the data is not corrupted during transmission. The verification formula is: ,in, This represents the check value. Indicates the first The content of each data block Indicates the total number of data blocks. This represents the modulus cardinality. The predictive data forwarding module records data reception logs, including reception time, data volume, and verification results, providing a basis for subsequent troubleshooting.

[0055] 2) The forecast data forwarding module performs data format conversion and standardization. Based on the requirements of the power grid dispatch center's forecast master station, the forecast data forwarding module converts its internal data format to the standard power system data format. The module converts timestamps to the standard time format required by the forecast master station, typically using the international standard time format while considering time zone offsets. The module converts power generation values ​​to standard units of measurement, usually kilowatts or megawatts, using the following conversion formula: ,in, This represents the converted standard power value. Indicates the original power value. This represents the unit conversion factor. The forecast data forwarding module organizes the data according to the data structure requirements of the forecast master station, including fields such as power plant identifier, forecast type, time range, and power value sequence.

[0056] 3) The predictive data forwarding module encapsulates the communication protocol and adds security identifiers. Based on the interface specifications of the predictive master station, the predictive data forwarding module selects an appropriate power system communication protocol, typically IEC-104 or IEC-102. The module encapsulates the standardized data according to the selected protocol, adding necessary fields such as protocol headers, control fields, and information bodies. It adds security authentication information to the data packets, including digital signatures and power plant identification, to ensure the trustworthiness of data transmission. Finally, the module uses encryption algorithms to protect sensitive data, preventing it from being stolen or tampered with during transmission.

[0057] 4) The predictive data forwarding module establishes a dedicated power channel connection. Through the network equipment in Security Zone II, the predictive data forwarding module establishes a stable communication connection with the power grid dispatch center's predictive master station. The predictive data forwarding module uses a dedicated power channel, a physically or logically isolated dedicated communication network with high reliability and security. The predictive data forwarding module executes a connection handshake protocol, performing identity verification and session negotiation with the predictive master station to ensure the legitimacy of the connection establishment. The predictive data forwarding module maintains connection status monitoring, detects network connectivity in real time, and automatically initiates a reconnection mechanism in the event of a connection interruption.

[0058] 5) The predictive data forwarding module transmits data and processes responses. The predictive data forwarding module sends the encapsulated data to the predictive master station through the established dedicated power channel. It employs a reliable transmission mechanism to ensure complete data packet delivery and implements a confirmation and retransmission strategy for critical data. The predictive data forwarding module receives response messages from the predictive master station and parses the response codes to determine if data transmission was successful. Based on the response results, the predictive data forwarding module updates its local transmission status record, arranging retransmission or recording anomalies for failed data transmissions. The predictive data forwarding module monitors data transmission performance in real time, statistically analyzing metrics such as transmission success rate and average latency.

[0059] 6) The predictive data forwarding module logs and generates reports. It records detailed information for each data transmission, including transmission time, data volume, response result, and exception information. The module periodically generates data transmission statistics reports, including success rate analysis, latency statistics, and exception classification. It provides a data transmission status query interface, allowing system administrators to monitor data uploads in real time. The module also has an alarm mechanism to promptly notify maintenance personnel in case of continuous transmission failures or abnormal performance indicators.

[0060] Security Zone I is the secure area within the power monitoring system for real-time control operations, also known as the production control zone. Security Zone I houses control systems and equipment directly involved in power production, including critical facilities such as photovoltaic power control devices and local photovoltaic monitoring devices. Security Zone I is connected to Security Zone II via a forward isolation device to ensure unidirectional data transmission. Systems within Security Zone I require the highest level of security and real-time performance, and all operations must comply with the power monitoring system's security protection regulations.

[0061] Security Zone II is the security zone within the power monitoring system for non-real-time control operations, also known as the management information zone. Security Zone II houses business systems that do not directly participate in real-time control, including distributed photovoltaic power generation prediction devices and data storage systems. Security Zone II is isolated from Security Zone I by a firewall and connected to the Internet zone through a reverse isolation device. Security Zone II allows non-real-time operations such as data processing and analysis, but must adhere to strict security access control policies.

[0062] The Internet Zone is a secure area within the power system that connects to the public network; it is also known as the external network zone. The Internet Zone houses systems that need to interact with the external network, including weather forecast servers and external data interfaces. The Internet Zone connects to the public internet via firewalls and exchanges data with Security Zone II through reverse isolation devices. The Internet Zone employs stringent security measures to prevent external network threats from spreading into the internal area.

[0063] The reverse isolation device is a crucial component of power system security, enabling unidirectional data transmission from lower-security areas to higher-security areas. Employing protocol stripping and data transfer technologies, it only allows data to flow unidirectionally from the Internet zone to Security Zone II. Through data caching and content inspection mechanisms, the reverse isolation device ensures the integrity and security of transmitted data. It is a key part of the power system's defense-in-depth cybersecurity system.

[0064] The BeiDou time synchronization device is a specialized unit that utilizes the BeiDou satellite navigation system to provide high-precision time synchronization signals. By receiving time synchronization signals transmitted by BeiDou satellites, the device provides a unified standard time source for all equipment in the power system. It supports multiple time synchronization methods, including network time protocols and second pulses, ensuring time consistency across all devices within the system. The BeiDou time synchronization device plays a crucial role in power systems for event sequence recording and fault analysis.

[0065] The forecasting master station is the core system of the power grid dispatch center, used to receive and process power generation forecast data. It aggregates power generation forecast data from various distributed photovoltaic power plants, performs data analysis and integration. The forecasting master station provides decision support to the power grid dispatching department, including functions such as power balance analysis and reserve capacity calculation. The forecasting master station establishes a secure and reliable data communication connection with the forecasting systems of each power plant through a dedicated power channel.

[0066] Currently, distributed photovoltaic (PV) power generation forecast data cannot be uploaded to the power grid dispatch center, preventing dispatch departments from obtaining PV output curves in advance and affecting power balance management. This invention effectively solves this technical problem through a forecast data forwarding module. This module converts the ultra-short-term, short-term, and medium-term PV power generation forecast data generated by the power generation forecast module into the standard data format required by the power grid dispatch center, ensuring data standardization and uniformity. The processed forecast data is reliably uploaded to the forecast master station through a dedicated power channel, ensuring the stability and security of data transmission. This design improves the data upload capability of distributed PV power stations, enabling dispatch departments to obtain accurate PV power generation forecast information in a timely manner. Simultaneously, the standardized data format improves system compatibility, and the dedicated power channel ensures communication quality, effectively avoiding dispatch difficulties caused by data transmission interruptions or format mismatches. This technical feature enhances the power grid's predictability of distributed PV output, provides reliable data support for optimizing power resource allocation, helps reduce spinning reserve capacity, and improves power grid operating efficiency.

[0067] Optionally, in the above technical solution, the distributed photovoltaic power generation prediction device further includes an application function module. This module is used to: display distributed photovoltaic power station information, real-time power generation data and power generation prediction trends, real-time meteorological data and weather forecasts, prediction accuracy indicators, and automatically generate prediction reports. Specifically: 1) The application function module obtains distributed photovoltaic power station information from the database module, including the installed capacity of the power station. Geographical coordinates And equipment configuration parameters. The application function module receives processed real-time power generation data from the data governance module, including the current power generation capacity. and cumulative power generation The application function module obtains power generation forecast trend data from the power generation forecast module, including ultra-short-term forecast sequences. Short-term forecast sequence and medium-term forecast sequence The application's functional modules also integrate real-time meteorological data. and weather forecast data This forms a complete dataset for display.

[0068] 2) The application functional modules utilize a Web technology stack to construct an interactive display interface, including functional areas such as a power plant overview area, a real-time monitoring area, a forecast display area, and a statistical analysis area. The application functional modules use visualization libraries such as ECharts or D3.js to configure various chart components, including line charts to display power generation trends, dashboards to display key operating parameters, maps to display the power plant's geographical location, and heat maps to display meteorological data distribution. The application functional modules are designed with a responsive layout to ensure that the display interface can be displayed correctly on various display devices, including the control room screen, desktop workstations, and mobile terminals.

[0069] 3) The application's functional modules implement real-time data updates and dynamic rendering mechanisms. These modules establish WebSocket connections to communicate with the backend service, receiving push updates of real-time power generation data and real-time meteorological data. The application's functional modules also set the data refresh cycle. For real-time power generation data, short-cycle updates (e.g., once every 5 seconds) are used, while for predicted trend data, long-cycle updates (e.g., once every 15 minutes) are used. The application module implements a dynamic chart rendering algorithm, smoothly updating the visualization as new data arrives, avoiding interface flickering or lag. The application module also provides a manual refresh control, allowing users to update the displayed content as needed.

[0070] 4) The application module calculates and displays prediction accuracy indicators. The application module retrieves historical prediction data from the database module. and corresponding actual power generation data Calculate the various prediction accuracy indicators. The root mean square error calculation formula is: ,in, This indicates the number of data points involved in the calculation. The formula for calculating the mean absolute percentage error is: The application's functional modules display the calculated prediction accuracy indicators in the form of numerical cards, trend charts, and radar charts, and support viewing accuracy changes at different time scales (day, week, month).

[0071] 5) The application module defines a standard template for forecast reports, including a fixed structure such as a title area, data summary area, chart area, and analysis description area. The application module sets report generation triggers, supporting both timed triggers (e.g., generating a daily report at 08:00) and event triggers (e.g., generating a monthly report at the end of the month). The application module collects the necessary report content from various data sources, including forecast results, actual data, meteorological records, and accuracy indicators, and populates them into the corresponding positions in the template. The application module uses a report generation engine to convert the populated template into a PDF document or Excel spreadsheet format, stores it in a specified directory, and automatically distributes it to relevant personnel via email.

[0072] 6) The application module implements a time range selector, allowing users to customize the data display for any time period. It provides data drill-down functionality, supporting detailed data viewing from the power plant level down to the generator unit level. The module also implements chart data export functionality, allowing users to export displayed chart data as CSV or JSON format. Finally, the module records user operation logs to facilitate tracking data access behavior and optimize user experience.

[0073] Distributed photovoltaic (PV) power station information refers to a set of static data describing the basic attributes and configuration parameters of a PV power station. This information includes key parameters such as power station name, installed capacity, geographical coordinates, component type, inverter specifications, grid connection point information, and commissioning date. Distributed PV power station information constitutes the digital identity of the power station, providing necessary background data and calculation benchmarks for power generation prediction. In system demonstrations, distributed PV power station information is typically presented in the form of a summary panel or configuration page, helping operation and maintenance personnel quickly understand the basic situation of the power station.

[0074] Real-time power generation data refers to the continuous collection of current power generation operating parameters from distributed photovoltaic power stations, including measured values ​​such as real-time power generation, cumulative power generation, voltage and current values, and equipment status. Power generation forecast trends refer to the prediction results of future power generation based on historical data and meteorological conditions, including power change curves formed by ultra-short-term, short-term, and medium-term forecasts. When real-time power generation data and power generation forecast trends are displayed together, they can intuitively reflect the current operating status and future power generation capacity of the power station, providing complete time-dimensional information for operation and maintenance decisions.

[0075] Real-time meteorological data refers to current environmental parameters collected by local meteorological monitoring equipment, including measured values ​​of light intensity, ambient temperature, relative humidity, wind speed and direction, and atmospheric pressure. Weather forecasts refer to predictions of future weather conditions obtained from weather forecast servers, including forecast values ​​of meteorological elements at different time scales and their changing trends. Comparing real-time meteorological data with weather forecasts helps assess forecast accuracy and understand the impact of meteorological conditions on power generation.

[0076] Among them, prediction accuracy indicators are a set of evaluation parameters used to quantify the accuracy of power generation prediction. These indicators include statistical quantities such as root mean square error, mean absolute error, mean absolute percentage error, and correlation coefficient. Prediction accuracy indicators objectively reflect the performance of the prediction model by systematically comparing predicted and actual values. In the application functional module, prediction accuracy indicators are displayed in numerical and visual form, providing a basis for model optimization and the evaluation of the reliability of prediction results.

[0077] The forecast report is a standardized document automatically generated by the system, summarizing power generation forecast data and analysis results for a specific time period. The forecast report includes forecast results, actual power generation data, accuracy assessment indicators, meteorological condition records, and explanations of any anomalies. The forecast report is generated according to a fixed period or trigger conditions, supporting different time granularities such as daily, weekly, and monthly reports. The forecast report is output in a structured format for easy archiving, distribution, and further analysis.

[0078] Currently, distributed photovoltaic (PV) power plants suffer from low observability and inefficient operation and maintenance (O&M) management. Operational data is scattered and lacks intuitive visualization, impacting O&M decision-making efficiency. This invention effectively improves this situation through an application module. This module centrally displays distributed PV power plant information, real-time power generation data and forecast trends, real-time meteorological data and forecasts, and forecast accuracy indicators, providing comprehensive operational status visualization. The automatic generation of forecast reports replaces traditional manual report creation, significantly improving data processing efficiency. This centralized display method enables O&M personnel to quickly grasp the power plant's operational status and future power generation trends, and promptly identify anomalies. Continuous display of forecast accuracy indicators provides a basis for model optimization, helping to improve long-term forecast accuracy. Simultaneously, standardized report formats ensure the standardization and comparability of data records, supporting historical data retrospective analysis. This technical feature significantly improves the O&M management efficiency of distributed PV power plants, providing an effective tool for optimized operation and refined management.

[0079] Optionally, in the above technical solution, the distributed photovoltaic power generation prediction device also includes a database module, which is used to store the ultra-short-term, short-term and medium-term power generation prediction results generated by the power generation prediction module, and supports retrospective query of historical data.

[0080] Current distributed photovoltaic (PV) power plants suffer from low observability and a lack of effective management of historical forecast data. This invention addresses this issue by storing ultra-short-term, short-term, and medium-term power generation forecasts generated by the power generation forecasting module through a database module, establishing a complete data archive. It supports retrospective queries of historical forecast data, providing data support for forecast model optimization and facilitating the analysis of patterns in forecast accuracy changes. This systematic data management approach enhances the usability of power plant operation data, laying the foundation for continuous improvement in power generation forecast accuracy.

[0081] Optionally, the above technical solution also includes a photovoltaic power control device, which is located in security zone I. The first firewall is connected to the photovoltaic power control device, and the zone I station control switch is connected to the photovoltaic power control device. The photovoltaic power control device is used to collect real-time power generation data and operating status of each power generation unit of the distributed photovoltaic power station, and to intelligently and flexibly regulate the power generation of the distributed photovoltaic power station based on the predicted data and trends of the power generation prediction module. Specifically: 1) The photovoltaic power control device collects real-time power generation data from each power generation unit. The photovoltaic power control device uses the monitoring network of the distributed photovoltaic power station to collect data at a fixed sampling period. Real-time power generation data is collected from each power generation unit. This data includes instantaneous power generation values. Voltage value Current value and frequency value The photovoltaic power control device establishes a connection with the power generation unit controller using a standard communication protocol, and reads data registers via Modbus TCP or IEC-61850 protocol. The photovoltaic power control device performs preliminary processing on the collected real-time power generation data, including unit standardization and data buffering, to provide standardized input for subsequent analysis.

[0082] 2) The photovoltaic power control device monitors the operating status of each power generation unit. The photovoltaic power control device obtains the operating status information of each power generation unit through a status monitoring interface, including the inverter operating mode. Grid connection status and protection device operation status The photovoltaic power control device analyzes status data to identify abnormal operating conditions of the power generation units, such as over-temperature alarms, insulation faults, and communication interruptions. The photovoltaic power control device then establishes a status matrix for the power generation units. It records the health status and availability of each power generation unit, providing a state basis for control decisions.

[0083] 3) The photovoltaic power control device receives prediction data from the power generation prediction module. The photovoltaic power control device obtains ultra-short-term power generation prediction data from the power generation prediction module through the firewall between Security Zone I and Security Zone II. Short-term power generation forecast data and medium-term power generation forecast data The photovoltaic power control device verifies the completeness and timeliness of the forecast data, checking whether the deviation between the timestamp and the current time is within the allowable range. The photovoltaic power control device stores the forecast data in an internal buffer and marks the data update time. .

[0084] 4) The photovoltaic power control device analyzes the trend of power generation changes. Based on the received forecast data, the photovoltaic power control device calculates the trend index of power generation changes. The photovoltaic power control device uses a linear regression method to analyze the slope of the ultra-short-term forecast data. The formula is: ,in, This indicates the number of data points used for trend analysis. Indicates the first At a certain point in time, Indicates the first The ultra-short-term predicted power values ​​at specific time points. The photovoltaic power control device determines the direction and intensity of power generation trends based on the sign and magnitude of the slope value, providing a basis for control strategies.

[0085] 5) The photovoltaic power control device formulates an intelligent and flexible regulation strategy. Combining real-time power generation data, operating status, forecast data, and trends, the photovoltaic power control device generates control commands according to a predetermined regulation logic. The photovoltaic power control device compares the current power generation... With predicted power Relationship, assess grid demand and regulatory capacity The photovoltaic power control device employs different control strategies based on five typical scenarios: ① When the predicted power is high and shows an upward trend, the photovoltaic power control device calculates the power limit value according to the grid dispatch requirements. Power reduction instructions are issued according to the priority order of power generation units to avoid power generation exceeding the grid demand.

[0086] ② When the predicted power is high and shows a downward trend, the photovoltaic power control device maintains the current operating state, does not issue control commands, records the operating data of each power generation unit generating power at maximum power, and automatically generates equipment inspection notices to remind on-site operation and maintenance personnel to check for potential hot spots and shading hazards.

[0087] ③ When the predicted power is relatively stable, the photovoltaic power control device makes fine adjustments according to the grid dispatch requirements. This allows for small-scale power adjustments to specific power generation units, improving power generation efficiency and economic benefits.

[0088] ④ When the predicted power is low and shows an upward trend, the photovoltaic power control device calculates the power increase. The output power of the power generation units will be gradually increased to meet the upcoming increase in grid demand in advance.

[0089] ⑤ When the predicted power is low and shows a downward trend, the photovoltaic power control device calculates the power limit value. Appropriately limit the output of power generation units to prevent power generation from falling below the grid demand.

[0090] 6) The photovoltaic power control device sends control commands, including power setpoints, to each power generation unit through the control network. , Run mode commands and state switching commands The photovoltaic power control device monitors the execution of control commands and verifies the actual power generation. With target power Is the deviation within the allowable error range? Inside. The photovoltaic power control device records key parameters during the regulation process, including response time. Adjustment accuracy Based on changes in equipment status, an evaluation report on the control effect is generated.

[0091] Currently, distributed photovoltaic (PV) power plants suffer from unstable power output, affecting grid balance. This invention uses a PV power control device to collect real-time data from each power generation unit and, based on the prediction results and trends from a power generation prediction module, achieves intelligent and flexible control of power generation. This control method dynamically adjusts the power plant output according to the predicted power level, ensuring stable grid operation while improving power generation efficiency and effectively mitigating the impact of distributed PV power output fluctuations on the power system.

[0092] Optionally, in the above technical solution, the data governance module uses statistical methods and business logic to filter out outliers in the real-time power generation data and meteorological data of distributed photovoltaic power stations, and further analyzes the outliers to remove them and noise, replacing them with reasonable values. Specifically: 1) The data governance module receives data and initializes processing parameters. The data governance module acquires real-time power generation data and meteorological data from the distributed photovoltaic power station from the data acquisition module. This data is transmitted in time-series format, including timestamps and corresponding data values. The data governance module sets the sliding window size. Thirty consecutive data points are used for localized data analysis. The data governance module loads a predefined business logic rule base, including the rated installed capacity of distributed photovoltaic power stations. Maximum allowable power change rate And reasonable range thresholds for various meteorological parameters. The data governance module initializes statistical detection parameters, including Z-score thresholds. Set to 3, moving standard deviation coefficient Set it to 2.

[0093] 2) The data governance module performs outlier detection based on statistical methods. For each data point, the data governance module calculates its statistical characteristics within a sliding window. For power generation data, the data governance module calculates the mean of the data within the window. and standard deviation Then calculate the Z-score value for each data point: ,in, This represents the value of the current power generation data point. If The data governance module marks this data point as a statistical outlier. For light intensity data in meteorological data, the data governance module uses the same method to calculate the mean of the light intensity data. and standard deviation The data governance module detects data points that deviate from the normal distribution. It also uses the moving standard deviation method to mark data points that differ excessively from the local standard deviation.

[0094] 3) The data governance module uses business logic rules to screen for outliers. Based on the operational characteristics of distributed photovoltaic power plants, the data governance module performs multiple business rule checks. For power generation data, the data governance module verifies whether each data point meets the requirements. Data points outside this range are marked as anomalies. The data governance module calculates the power change rate between adjacent time points: ,in, This represents the power generation value at the current time. This represents the power generation value at the previous time point. Indicates the sampling time interval. If... The data governance module marks the data point as an anomaly. For meteorological data, the data governance module checks whether each parameter is within a reasonable physical range, such as whether the light intensity is between 0-1500W / m² and the temperature is between -50°C and 60°C. Data points that exceed these ranges are marked as anomalies.

[0095] 4) The data governance module distinguishes between noise points and true outliers. It further analyzes the marked outliers, using a moving average filtering method to identify high-frequency noise. The data governance module calculates the moving average of the data: ,in, This represents the smoothed power value. Indicates the first in the window The values ​​of each power data point. The data governance module calculates the deviation between the raw data and the smoothed data. ,if Less than the noise threshold If an outlier is identified, the data point is classified as noise; otherwise, it is classified as a true anomaly. The data governance module analyzes the distribution pattern of outliers, identifies isolated outliers and continuous outlier segments, and provides a basis for different processing strategies.

[0096] 5) The data governance module performs outlier handling and data repair. The data governance module takes appropriate measures based on the type of outlier. For genuine outliers that clearly do not conform to business logic, such as negative power generation values ​​or meteorological data values ​​exceeding physical limits, the data governance module performs a deletion operation, removing these data points from the data sequence. For repairable outliers, the data governance module uses linear interpolation to calculate reasonable replacement values. ,in, This indicates the power value after the replacement. This indicates the previous normal power value. This indicates the next normal power value. Indicates the current time point, Indicates the previous point in time. This indicates the next time point. When multiple consecutive data points are abnormal, the data governance module replaces them with the mean of the normal data within the window. ,in, This represents the replacement value for the mean. This indicates the number of normal data points within the window. Indicates the first The values ​​for each normal data point are used. For noisy data points, the data governance module applies a filtering algorithm to smooth them.

[0097] 6) The data governance module verifies the processing results and outputs high-quality data. The data governance module performs quality checks on the processed data to ensure that all data points are within a reasonable range and that the time series is continuous. The data governance module calculates data quality indicators, including the anomaly rate. and processing efficiency The data governance module evaluates the effectiveness of the treatment. It sends the treated real-time power generation data and meteorological data from the distributed photovoltaic power station to subsequent modules, including the meteorological forecast data analysis module and the power generation prediction module. The data governance module generates a data governance report, recording detailed logs of anomaly detection and handling, including the number of anomalies, handling methods, and quality indicators, for system monitoring and optimization.

[0098] In another embodiment, the system includes: a BeiDou time synchronization device, a photovoltaic power control device, a photovoltaic local monitoring device, and a zone I station control switch, all located in Security Zone I; a firewall connecting the devices in Security Zone I and Security Zone II; a distributed photovoltaic power generation prediction device and a zone II switch, all located in Security Zone II; a reverse isolation device connecting the devices in Security Zone II and the Internet zone; and a weather forecast server and a firewall, all located in the Internet zone. Specifically, the zone I station control switch connects the BeiDou time synchronization device, photovoltaic power control device, and photovoltaic local monitoring device located in Security Zone I; the zone II switch connects the distributed photovoltaic power generation prediction device and the reverse isolation device; the weather forecast server obtains weather forecast data from multiple weather forecast websites via the firewall and transmits the weather forecast data to the reverse isolation device; the BeiDou time synchronization device provides time synchronization services for the devices in Security Zone I; the photovoltaic local monitoring device monitors the operating status of the distributed photovoltaic power station and collects real-time power generation data; and the photovoltaic power control device collects real-time power generation data and operating status of each power generation unit in the distributed photovoltaic power station and performs intelligent and flexible control based on power generation prediction data and trends.

[0099] The distributed photovoltaic (PV) power generation prediction device includes a data acquisition module, a data governance module, a meteorological forecast data analysis module, a power generation prediction module, an application function module, a prediction data forwarding module, and a database module. The data acquisition module can obtain real-time and historical power generation data from the PV local monitoring device or SCADA system in the power plant's safety zone I via a firewall, using standard communication protocols such as ModbusTCP, IEC-101 / 103 / 104, MQTT, DL / T645, and IEC-61850. The data governance module can collect and preprocess the real-time and historical power generation and meteorological data obtained from the power plant's safety zone I. First, statistical methods and business logic are used to filter out outliers. Then, the outliers are further analyzed to remove them and noise, and replaced with reasonable values ​​to improve data quality. Statistical methods include calculating the Z-score and moving standard deviation of data points, and business logic includes checking whether the data is within the rated installed capacity range or conforms to reasonable meteorological parameter ranges. The weather forecast data analysis module can receive weather forecast data from the reverse isolation device and perform accuracy analysis on weather forecast data from the weather forecast server and processed weather forecast data from the data governance module to determine the weather forecast data with high accuracy. The weather forecast data analysis module can automatically acquire key meteorological elements such as light intensity, temperature, humidity, wind speed, and wind direction from multiple weather forecast websites, and regularly compare and analyze them with real-time power generation data from distributed photovoltaic power stations and real-time meteorological data collected by local weather instruments. It calculates the prediction error of each meteorological element, performs correlation analysis between the actual changes of meteorological elements and the changes in power generation data of the power station, and finally, based on the comparative analysis results, automatically selects the weather forecast data source with better forecast accuracy as the input for the weather forecast data in the next period. The power generation prediction module can predict the ultra-short-term, short-term, and medium-term power generation of distributed photovoltaic (PV) power plants based on high-precision meteorological forecast data and processed power generation data output by the data governance module. Ultra-short-term power generation prediction covers a timescale of 15 minutes to 4 hours, short-term power generation prediction covers a timescale of 1 day to 3 days, and medium-term power generation prediction covers a timescale of 4 days to 10 days. The module uses deep learning models for modeling and prediction, including a long short-term memory (LSTM) network architecture, to process time-series data and capture long-term dependencies. The application function module can display distributed PV power plant information, real-time power generation data and prediction trends, real-time meteorological data and forecasts, and prediction accuracy indicators on a large screen, and automatically generate relevant prediction reports, improving the operation and maintenance efficiency of distributed PV power plants. Display methods include line charts, dashboards, and map visualizations, and report generation supports daily, weekly, and monthly reports.The forecast data forwarding module converts the ultra-short-term, short-term, and medium-term power generation forecast data generated by the power generation forecasting module into the data format required by the power grid dispatch center, and uploads it to the forecasting master station through a dedicated power channel. The module supports IEC-104 or IEC-102 protocols for data encapsulation and transmission, ensuring data security and reliability. The database module stores various forecast data generated by the power generation forecasting module, including ultra-short-term, short-term, and medium-term power generation forecast results, and supports users in retrospectively querying stored historical data. Simultaneously, the time-series database efficiently processes time-series data and supports real-time monitoring and analysis. The database module also records log information during data governance and forecasting processes for system optimization and troubleshooting.

[0100] The photovoltaic power control device has the following functions: ① Collect real-time power generation data and operating status of each power generation unit in the distributed photovoltaic power station. Real-time power generation data includes power generation, voltage, current, and frequency. Operating status includes inverter operating mode, grid connection status, and fault codes. ② Obtain power generation prediction data from the distributed photovoltaic power generation prediction device and intelligently and flexibly regulate the power generation of the distributed photovoltaic power station based on the predicted power magnitude and trend. Specific regulation strategies include: when the predicted power is high and trending upward, appropriately limiting the output of the power generation units according to grid dispatch requirements to avoid exceeding grid demand; the limiting method includes issuing power reduction commands to the power generation unit controller. When the predicted power is high and trending downward, no regulation commands are issued, allowing each power generation unit to generate power at maximum power, and automatically reminding on-site maintenance personnel to check for potential hot spots, shading, and other hidden dangers; reminders are sent via system notification or email. When the predicted power is relatively stable, fine-tuning is performed according to grid dispatch requirements to improve power generation efficiency and economic benefits; fine-tuning includes small adjustments to the output power of the power generation units. When the predicted power is low but shows an upward trend, the output power of the power generation units will be appropriately increased to meet the upcoming increase in grid demand in advance; this increase will be achieved by gradually raising the power setpoint of the power generation units. When the predicted power is low but shows a downward trend, the output of the power generation units will be appropriately limited to prevent the generated power from falling below grid demand; this will limit the power limit value calculated based on the predicted data and issue control commands. The photovoltaic power control device also monitors the execution effect of the control commands, verifies the deviation between the actual generated power and the target power, and records the control process parameters for evaluation and optimization.

[0101] To address the challenges of low observability, high uncertainty in meteorological conditions, poor communication quality, and inability to upload forecast data to the power grid dispatch center in distributed photovoltaic (PV) power plants, this paper proposes a power generation forecasting system suitable for distributed PV power plants. This system enables distributed PV power plants to achieve high-precision power generation forecasting across multiple time scales and uploads the forecast data to the power grid dispatch center's main forecasting station. Specifically, it includes: a BeiDou time synchronization device, a PV power control device, a PV local monitoring device, and a zone I station control switch located in Security Zone I; a firewall connecting the devices in Security Zone I and Security Zone II; a distributed PV power generation forecasting device and a zone II switch located in Security Zone II; a reverse isolation device connecting the devices in Security Zone II to the internet area; and a weather forecast server and a firewall located in the internet area. Specifically, the zone I station control switch connects the BeiDou time synchronization device, PV power control device, and PV local monitoring device located in Security Zone I; the zone II switch connects the distributed PV power generation forecasting device to the reverse isolation device; and the weather forecast server obtains weather forecast data from weather forecast websites via the firewall and transmits it to the reverse isolation device.

[0102] Optionally, in the above technical solution, the weather forecast server set up in the Internet region obtains weather forecast data from different weather websites through the Internet region. Then, through the Internet region firewall and the security zone II reverse isolation device, the weather forecast data is first converted into an E text file and transmitted to the reverse isolation device. Then, the distributed photovoltaic power generation prediction device receives the E text file from the reverse isolation device and converts it into a standard data format, and transmits it to the power generation prediction module for use.

[0103] Optionally, in the above technical solution, the distributed photovoltaic power generation prediction device obtains real-time power generation data and meteorological data, historical power generation data and meteorological data of the distributed photovoltaic power station from the local photovoltaic monitoring device or SCADA system through the firewall of Security Zone I, and transmits the obtained data to the data governance module for processing. The processed data is then provided to the power generation prediction module for use.

[0104] Optionally, in the above technical solution, when the distributed photovoltaic power generation prediction device connects to the real-time and historical data of the distributed photovoltaic power station in Safety Zone I, it supports standard communication protocols such as Modbus TCP, IEC-101 / 103 / 104, MQTT, DL / T645, and IEC-61850, and also supports customized communication protocol development to ensure that the system has strong device access capabilities and data acquisition capabilities.

[0105] Optionally, in the above technical solution, when the distributed photovoltaic power generation prediction device interacts with the prediction master station of the power grid dispatch center through the prediction data forwarding module, it can encapsulate the prediction data in accordance with the IEC-104 protocol or the IEC-102 protocol and upload it to the prediction master station through a dedicated power channel to ensure the security and reliability of data transmission.

[0106] Optionally, in the above technical solution, the data governance module in the distributed photovoltaic power generation prediction device uses statistical methods and business logic to filter out outliers in the real-time power generation data and meteorological data of the distributed photovoltaic power station, and further analyzes the outliers to remove outliers and noise and replace them with reasonable values. Specifically, this includes calculating the Z-score value of the data points, checking whether the data is within the rated installed power range, verifying whether the meteorological parameters are within a reasonable range, and repairing the data using linear interpolation or mean replacement methods.

[0107] Optionally, in the above technical solution, the meteorological forecast data analysis module in the distributed photovoltaic power generation prediction device can receive meteorological forecast data from the reverse isolation device, and perform accuracy analysis on the meteorological forecast data from the meteorological forecast server and the processed meteorological forecast data from the data governance module. By calculating the prediction error of each meteorological element and performing correlation analysis between the actual changes of meteorological elements and the changes of power generation data of the station, the system automatically selects the meteorological forecast data source with better forecast accuracy as the input of meteorological forecast data for the next period of time.

[0108] Optionally, in the above technical solution, the power generation prediction module in the distributed photovoltaic power generation prediction device uses a deep learning model to perform ultra-short-term power generation prediction, short-term power generation prediction, and medium-term power generation prediction for the distributed photovoltaic power station based on high-precision meteorological forecast data and processed power generation data output by the data governance module. The ultra-short-term power generation prediction covers a time scale of 15 minutes to 4 hours, the short-term power generation prediction covers a time scale of 1 day to 3 days, and the medium-term power generation prediction covers a time scale of 4 days to 10 days.

[0109] Optionally, in the above technical solution, the photovoltaic power control device collects real-time power generation data and operating status of each power generation unit of the distributed photovoltaic power station, and intelligently and flexibly regulates the power generation of the distributed photovoltaic power station based on the predicted data and changing trends of the power generation prediction module. According to the predicted power size and changing trends, the device executes corresponding regulation strategies, including limiting the output of the power generation unit, maintaining maximum power generation, fine-tuning, increasing the output power, and other regulation methods.

[0110] like Figure 2As shown, the system includes a BeiDou time synchronization device, a photovoltaic power control device, a photovoltaic local monitoring device, and a station control switch in Zone I, all located in Security Zone I. The BeiDou time synchronization device provides high-precision time synchronization services for all equipment in Security Zone I, ensuring consistent time references for all system components. The station control switch in Zone I connects to the photovoltaic local monitoring device and photovoltaic power control device deployed in Security Zone I, and establishes a secure connection with the Zone II switch located in Security Zone II via a firewall. The Zone II switch in Security Zone II connects to the distributed photovoltaic power generation forecasting device deployed in Security Zone II, forming the core processing unit of the system. The weather forecast server in the Internet region is connected to the Zone II switch in Security Zone II via a reverse isolation device, and simultaneously connected to the Internet via a firewall, enabling secure communication with external meteorological data sources. The distributed photovoltaic power generation forecasting device is connected to the forecasting master station of the power grid dispatch center via a dedicated power channel, ensuring secure and reliable transmission of forecast data. The distributed photovoltaic power generation forecasting device obtains real-time power generation data and meteorological data, as well as historical power generation data and meteorological data, from the photovoltaic local monitoring device or SCADA system deployed in Security Zone I via the firewall in Security Zone I. These data include power generation data collected from photovoltaic inverters and environmental monitoring data collected from weather instruments. The weather forecast server, deployed in the Internet zone, first obtains weather forecast data from different weather websites via a firewall, then converts it into E-text file format and transmits it to the reverse isolation device. The distributed photovoltaic power generation forecasting device, deployed in Security Zone II, receives the E-text file from the reverse isolation device, parses it through the weather forecast data interface module, completes the weather data conversion, and converts it into a standard weather forecast data format for system use. The distributed photovoltaic power generation forecasting device maintains a stable connection with the power grid dispatch center's forecasting master station through a dedicated power channel. When interacting with the power grid dispatch center's forecasting master station, the forecast data forwarding module can encapsulate and upload the forecast data using the IEC-104 or IEC-102 protocol, ensuring that data transmission complies with power industry standards. The photovoltaic power control device has the following functions: collecting real-time power generation data and real-time operating status of each power generation unit of the distributed photovoltaic power station, which comes from the photovoltaic inverter and related monitoring equipment; acquiring power generation prediction data from the distributed photovoltaic power generation prediction device, and intelligently and flexibly controlling the power generation of the distributed photovoltaic power station according to the predicted power magnitude and trend.Specific control strategies include: when the predicted power is high and trending upward, appropriately limiting the output of power generation units according to grid dispatch requirements; when the predicted power is high and trending downward, maintaining each power generation unit at maximum power and automatically reminding maintenance personnel to check for potential hazards; when the predicted power is relatively stable, making fine adjustments according to grid dispatch requirements; when the predicted power is low and trending upward, appropriately increasing the output power of power generation units; and when the predicted power is low and trending downward, appropriately limiting the output of power generation units.

[0111] like Figure 3 As shown, the distributed photovoltaic power generation prediction device includes: a data acquisition module that acquires real-time and historical power generation and meteorological data from the local photovoltaic monitoring device or local SCADA system via a secure zone I firewall. Data sources include field equipment such as photovoltaic inverters and weather instruments. A data governance module performs data preprocessing and anomaly cleaning on the acquired data, using statistical methods and business logic to filter out outliers. Data significantly exceeding reasonable limits is marked and processed, and data quality is ensured through linear interpolation or mean replacement. A meteorological forecast data interface module receives meteorological forecast E-text files from the reverse isolation device, performs E-text file parsing and meteorological data conversion, transforming the raw data into the system's standard format. A meteorological forecast data analysis module performs forecast data verification and prediction data comparison, periodically comparing and analyzing the forecast with actual power plant operating data, calculating prediction errors, and automatically selecting the optimal meteorological data source.

[0112] The power generation prediction module, through modeling of power generation equipment and the power plant, enables ultra-short-term, short-term, and medium-term power generation prediction. The prediction data forwarding module handles data format conversion and uploads the predicted data, sending the processed data to the power grid dispatch center. The application function module connects to a large display screen, enabling functions such as prediction data display, prediction report generation, event querying, and access control. The database module is responsible for storing prediction data, efficiently managing time-series data through a time-series database, and supporting historical data backtracking and real-time monitoring and analysis. Through the collaborative work of these modules, this invention achieves accurate prediction and intelligent control of distributed photovoltaic power generation. The system structure is simple and reliable, suitable for both newly built distributed photovoltaic power plant deployments and the system upgrades of existing distributed photovoltaic power plants.

[0113] The beneficial effects of this invention include: it has the ability to manage real-time and historical data, and can collect and preprocess real-time power generation data and meteorological data, as well as historical power generation data and meteorological data from distributed photovoltaic power stations obtained from Security Zone I. First, it uses statistical methods and business logic to filter out outliers, and then further analyzes the outliers to remove outliers and noise, and replaces them with reasonable values. This solves the problems of poor communication quality and data anomalies caused by low compatibility of communication equipment and aging sensors, thereby improving data quality and increasing the accuracy of power generation prediction. It possesses the capability to acquire and analyze meteorological forecast data from multiple sources. It can automatically acquire key meteorological elements such as light intensity, temperature, humidity, wind speed, and wind direction from multiple meteorological forecast websites, and periodically compare and analyze this data with real-time power generation data from distributed photovoltaic power stations and real-time meteorological data collected by local weather instruments. It calculates the prediction error of each meteorological element, correlates the actual changes in meteorological elements with changes in power generation data from the power station, and finally, based on the comparative analysis results, automatically selects the meteorological forecast data source with better forecast accuracy as the input for the next period. This solves the problem of distributed photovoltaic power stations lacking more accurate meteorological forecast data or only being able to obtain forecast data from a single meteorological source, thus improving the accuracy of power generation prediction. It also possesses the capability to predict photovoltaic power generation at multiple time scales. It can model distributed photovoltaic power stations and power generation equipment, and perform ultra-short-term power generation predictions of 15 minutes to 4 hours, short-term power generation predictions of 1 day to 3 days, and medium-term power generation predictions of 4 days to 10 days. The prediction accuracy at each time scale is high, solving the problem of distributed photovoltaic power stations being unable to effectively predict power generation. It possesses the capability to forward forecast data northward, converting ultra-short-term (15 minutes to 4 hours), short-term (1 day to 3 days), and medium-term (4 days to 10 days) power generation forecast data into the data format required by the power grid dispatch center's forecast master station. This data is then uploaded to the forecast master station via a dedicated power channel, resolving the issue of distributed photovoltaic (PV) power generation forecast data being unable to be uploaded. Furthermore, it exhibits intelligent and flexible power generation control capabilities, enabling intelligent regulation of distributed PV power generation based on forecast data and trends, according to grid dispatch requirements. This includes various control strategies such as output limiting, maintaining maximum power generation, and fine-tuning, improving the operational efficiency and grid compatibility of distributed PV power stations. In other words, this invention collects real-time power generation data and operational status from each power generation unit of a distributed PV power station; acquires power generation forecast data from the distributed PV power generation forecasting device; and intelligently and flexibly regulates the power generation of the distributed PV power station based on the predicted power magnitude and trends.Specific control strategies include: when the predicted power is high and trending upward, the output of power generation units will be appropriately limited according to grid dispatch requirements to avoid exceeding grid demand; when the predicted power is high and trending downward, no control commands will be issued, allowing each power generation unit to generate power at maximum capacity, and automatically reminding on-site maintenance personnel to check for potential hot spots, shading, and other hidden dangers; when the predicted power is relatively stable, fine-tuning will be made according to grid dispatch requirements to improve power generation efficiency and economic benefits; when the predicted power is low and trending upward, the output power of power generation units will be appropriately increased to meet the upcoming increase in grid demand in advance; when the predicted power is low and trending downward, the output of power generation units will be appropriately limited to prevent power generation from falling below grid demand. Furthermore, the photovoltaic power control device can automatically acquire photovoltaic output prediction data from distributed photovoltaic power generation prediction devices and perform forward-looking flexible control of photovoltaics based on the output prediction data, avoiding frequent approaches of grid operating parameters to their limits, making the output fluctuations of distributed photovoltaics smoother, and improving the grid-friendliness of distributed photovoltaics.

[0114] like Figure 4 As shown in the figure, an embodiment of the present invention provides a method for predicting the power generation of a distributed photovoltaic power station. The method employs any distributed photovoltaic power station power generation prediction system and includes the following steps: S1, acquiring weather forecast data from multiple weather forecast websites via a weather forecast server; S2, acquiring power generation data and weather forecast data of the distributed photovoltaic power station from a local photovoltaic monitoring device or SCADA system via a data acquisition module; S3, performing anomaly detection and processing on the power generation data and weather forecast data acquired by the data acquisition module via a data governance module; S4, receiving and performing accuracy analysis on the weather forecast data from the weather forecast server and the processed weather forecast data from the data governance module via a weather forecast data analysis module to determine the weather forecast data with high accuracy; S5, predicting the power generation of the distributed photovoltaic power station for ultra-short-term, short-term, and medium-term periods based on the high-accuracy weather forecast data and the processed power generation data output by the data governance module via a power generation prediction module.

[0115] It should be noted that the beneficial effects of the power generation prediction method for a distributed photovoltaic power station provided in the above embodiments are the same as those of the power generation prediction system for a distributed photovoltaic power station described above, and will not be repeated here. Furthermore, the method and system embodiments provided in the above embodiments belong to the same concept, and their specific implementation processes are detailed in the system embodiments, and will not be repeated here.

[0116] An electronic device according to an embodiment of the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the aforementioned method for predicting the power generation of a distributed photovoltaic power station. Another embodiment of the present invention includes a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the aforementioned method for predicting the power generation of a distributed photovoltaic power station.

[0117] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A power generation prediction system suitable for distributed photovoltaic power plants, characterized in that, include: Weather forecast server and distributed photovoltaic power generation prediction device containing data acquisition module, data governance module, weather forecast data analysis module and power generation prediction module; The weather forecast server is used to: obtain weather forecast data from multiple weather forecast websites; The data acquisition module is used to: acquire power generation data and weather forecast data of the distributed photovoltaic power station from the local photovoltaic monitoring device or SCADA system; The data governance module is used to: perform anomaly detection and processing on the power generation data and weather forecast data of the distributed photovoltaic power station acquired by the data acquisition module; The weather forecast data analysis module is used to: receive and perform accuracy analysis on weather forecast data from the weather forecast server and processed weather forecast data from the data governance module, and determine the weather forecast data with high accuracy; The power generation prediction module is used to predict the power generation of the distributed photovoltaic power station in the ultra-short term, short term, and medium term, based on high-precision meteorological forecast data and the processed power generation data output by the data governance module.

2. The power generation prediction system for distributed photovoltaic power plants according to claim 1, characterized in that, The distributed photovoltaic power generation prediction device is located in Security Zone II, the weather forecast server is located in the Internet Zone, and the system also includes: A Beidou time synchronization device, a photovoltaic local monitoring device, and a station control switch in Security Zone I are installed; a first firewall connects the Beidou time synchronization device, the photovoltaic local monitoring device, and the station control switch in Security Zone I to the equipment in Security Zone II; a switch in Security Zone II is installed in Security Zone II; a reverse isolation device connects the distributed photovoltaic power generation prediction device and the switch in Security Zone II to the Internet area; and a second firewall is installed in the Internet area. The BeiDou time synchronization device is used to provide time synchronization services for the equipment in the security zone I; the photovoltaic local monitoring device is used to monitor the operating status of the distributed photovoltaic power station and collect real-time power generation data; the zone I station control switch connects the BeiDou time synchronization device and the photovoltaic local monitoring device; the zone II switch connects the distributed photovoltaic power generation prediction device and the reverse isolation device; the weather forecast server is used to obtain weather forecast data from multiple weather forecast websites through the second firewall and transmit the weather forecast data to the reverse isolation device. The weather forecast data analysis module is specifically used to: receive and perform accuracy analysis on the weather forecast data from the reverse isolation device and the processed weather forecast data from the data governance module, and determine the weather forecast data with high accuracy.

3. A power generation prediction system suitable for distributed photovoltaic power stations according to claim 1 or 2, characterized in that, The distributed photovoltaic power generation prediction device also includes a prediction data forwarding module, which is used to convert the ultra-short-term, short-term and medium-term power generation prediction data generated by the power generation prediction module into the data format required by the power grid dispatch center, and upload it to the prediction master station through a dedicated power channel.

4. A power generation prediction system suitable for distributed photovoltaic power stations according to claim 1 or 2, characterized in that, The distributed photovoltaic power generation prediction device also includes an application function module, which is used to: display the distributed photovoltaic power station information, real-time power generation data and power generation prediction trends, real-time meteorological data and weather forecasts, prediction accuracy indicators, and automatically generate prediction reports.

5. A power generation prediction system suitable for distributed photovoltaic power stations according to claim 1 or 2, characterized in that, The distributed photovoltaic power generation prediction device also includes a database module, which stores the ultra-short-term, short-term and medium-term power generation prediction results generated by the power generation prediction module and supports backtracking queries of historical data.

6. A power generation prediction system suitable for distributed photovoltaic power plants according to claim 1 or 2, characterized in that, It also includes a photovoltaic power control device, which is located in the security zone I. The first firewall is connected to the photovoltaic power control device, and the station control switch in zone I is connected to the photovoltaic power control device. The photovoltaic power control device is used to collect real-time power generation data and operating status of each power generation unit of the distributed photovoltaic power station, and to perform intelligent and flexible regulation of the power generation of the distributed photovoltaic power station based on the prediction data and changing trends of the power generation prediction module.

7. A power generation prediction system suitable for distributed photovoltaic power stations according to claim 1 or 2, characterized in that, The data governance module uses statistical methods and business logic to filter out outliers in the real-time power generation data and meteorological data of the distributed photovoltaic power station, and further analyzes the outliers to remove outliers and noise and replace them with reasonable values.

8. A method for predicting the power generation of a distributed photovoltaic power station, characterized in that, The method of using a power generation prediction system suitable for distributed photovoltaic power plants according to any one of claims 1 to 7 includes: Weather forecast data is obtained from multiple weather forecast websites through the aforementioned weather forecast server; The data acquisition module obtains power generation data and weather forecast data of the distributed photovoltaic power station from the local photovoltaic monitoring device or SCADA system. The data governance module performs anomaly detection and processing on the power generation data and weather forecast data of the distributed photovoltaic power station acquired by the data acquisition module. The meteorological forecast data analysis module receives and performs accuracy analysis on meteorological forecast data from the meteorological forecast server and processed meteorological forecast data from the data governance module to determine meteorological forecast data with high accuracy. The power generation prediction module uses high-precision meteorological forecast data and processed power generation data output by the data governance module to predict the power generation of the distributed photovoltaic power station in the ultra-short term, short term, and medium term.

9. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the power generation prediction method for a distributed photovoltaic power station as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the power generation prediction method for a distributed photovoltaic power station as described in claim 8.

Images