Low-voltage distributed photovoltaic output prediction and group scheduling and group control method based on big data model

Abstract

The invention relates to the field of low-voltage distributed photovoltaic regulation and control, and aims to solve the problems caused by high-proportion access to a power distribution network. Historical output, weather and geographic information data are collected and preprocessed, then a power prediction model fusing physics and machine learning is constructed, model hyper-parameters are optimized through historical data, and prediction accuracy is improved. On the basis, photovoltaic output capacity analysis is carried out, power prediction and power grid load requirements are combined, a regulation and control strategy is formulated, and the regulation and control strategy comprises construction of a dynamic simulation model, early warning of reverse overload and the like. Through a group regulation and control scheme, regulation and control are reduced according to a user rated capacity proportion in an equal proportion, and output regulation of the inverter is realized through an electricity utilization information acquisition system and the like. According to the invention, photovoltaic output can be accurately predicted, effective regulation and control can be realized, power balance of the whole grid in a special period can be realized, power grid stability and clean energy utilization efficiency can be improved, and sustainable development of distributed photovoltaic in the power grid can be promoted.

Description

Technical Field

[0001] The present invention relates to the field of low-voltage distributed photovoltaic control technology, and in particular to a low-voltage distributed photovoltaic output prediction and group control method based on a big data model. Background Art

[0002] As global demand for clean energy continues to grow, distributed photovoltaics, thanks to their flexible installation methods and local consumption, are increasingly being integrated into distribution networks. However, this high integration rate is a double-edged sword: while it brings clean energy benefits, it also presents numerous challenges.

[0003] On the one hand, low-voltage distributed photovoltaic systems are characterized by their large number, extremely dispersed distribution, and small installed capacity per household. This makes unified and efficient management and control extremely difficult. Traditional grid management models are unable to adapt to this decentralized energy landscape, like dealing with a pile of loose sand, and are unable to cope with it.

[0004] On the other hand, it poses a series of severe challenges to the stable operation of the distribution network. The disorderly integration of a large number of distributed photovoltaic systems leads to frequent reverse overloads of distribution transformers, placing an excessive burden on the "heart" of the distribution network. 10kV reverse power flow disrupts the previously orderly flow of power, complicating the power distribution of the grid. High voltage at the end of distribution lines is a hidden danger that could affect power supply quality at any time and threaten the safe and stable operation of electrical equipment.

[0005] In the actual operation of the power grid, special periods (such as peak electricity demand and emergency power periods during extreme weather) place extremely high demands on the overall power balance of the entire network. These periods are often accompanied by rapid fluctuations in load. If the output of low-voltage distributed photovoltaic systems cannot be effectively managed, not only will the clean energy advantages of the system fail to contribute to grid stability, but the risk of imbalance will be further exacerbated.

[0006] Therefore, there is an urgent need for an innovative and effective method that can not only accurately grasp the output characteristics of low-voltage distributed photovoltaics, but also implement intelligent group adjustment and control to perfectly match them with the load demand of the power grid, thereby ensuring that the power grid can operate stably under any operating conditions and achieve the key goal of balancing the power of the entire network during special periods. Summary of the Invention

[0007] In order to solve the above technical problems, the present invention provides a low-voltage distributed photovoltaic power prediction and group regulation and control method based on a big data model. Through accurate output prediction and effective group regulation and control strategy, the stable operation of low-voltage distributed photovoltaics in the power grid is achieved, and it participates in the peak regulation of the power grid to achieve the goal of power balance of the entire network during special periods.

[0008] The present invention provides a method for predicting and controlling low-voltage distributed photovoltaic output based on a big data model, comprising the following steps:

[0009] Step 1: Data collection and analysis: Collect historical output data, meteorological data, and geographic information data of low-voltage distributed photovoltaic systems, and pre-process the collected data, including cleaning and normalization operations;

[0010] Step 2: Build a photovoltaic power prediction model: Based on the physical principles of photovoltaic power generation, consider various influencing factors to build a photovoltaic power generation prediction fusion model that integrates multiple machine learning regression methods. Use historical data to fit and correct the hyperparameters of the photovoltaic power generation prediction fusion model, including model research and selection, and model training.

[0011] Step 3: Developing a PV Control Strategy: Analyze PV output capacity, power forecast, and control demand to determine a PV control strategy. This strategy includes building a dynamic output capacity simulation model, using NWP data to predict power, analyzing control demand and time requirements, and developing a plan based on regional output capacity indicators. This strategy also includes providing a PV reverse overload warning.

[0012] Step 4: Implementation of group regulation and control scheme: Generate a list of regulated users according to the PV regulation strategy, and adjust the regulation in proportion to the user's rated capacity. Send control commands to the collection terminal through the electricity consumption information collection system, and then to the protocol converter, and finally adjust the inverter output to achieve group regulation and control.

[0013] Furthermore, the meteorological data includes solar radiation intensity, temperature, humidity, air pressure, wind speed and wind direction, and the geographic information data includes the latitude and longitude, altitude and topography of the power station; the cleaning operation includes removing obviously erroneous data points, and supplementing the missing data using linear interpolation or mean interpolation methods; the normalization operation includes mapping the data to the [0,1] interval, establishing a data interface with the distributed photovoltaic power station monitoring system, meteorological department and geographic information system, and obtaining real-time data, and the data acquisition time interval is every hour or shorter.

[0014] Furthermore, in the training and optimization of the photovoltaic power prediction fusion model, after model investigation and comparison of various types of models, a hybrid model that integrates the physical model and the machine learning method is selected. The model training includes data partitioning, training, verification and testing operations; in the model training process, cross-validation technology is used to select the optimal model hyperparameter combination, and the model is retrained regularly based on newly acquired historical data. At least one of the support vector machine, neural network, and decision tree is selected as the regression method in the fusion model; based on the physical principles of photovoltaic power generation, a multi-parameter physical model is established. The multi-parameters include angle, area, shading, irradiance and temperature. The parameters in the physical model are estimated and optimized through time series analysis methods to construct a photovoltaic power prediction model.

[0015] Furthermore, in the training and optimization of the photovoltaic power prediction model, the neural network models introduced are transformer and linear models. The bagging ensemble learning method is used to integrate multiple base models. The base models are trained on different sub-training sets obtained by random sampling with replacement of the training set.

[0016] Furthermore, in the formulation of photovoltaic control strategies, machine learning algorithms are used to establish a relationship model between photovoltaic output capacity, environmental factors, and equipment status. PV output capacity indicators under different conditions are calculated in real time to construct a dynamic output capacity simulation model. Power is predicted using NWP data. Specifically, the photovoltaic power output for future time periods is predicted based on numerical meteorological forecast data and a trained photovoltaic power generation prediction model. Control demand and time requirements are analyzed. Specifically, the amount of control power and control time requirements required by photovoltaic power stations in different time periods are analyzed in combination with the grid load demand forecast. A plan is formulated based on regional output capacity indicators. Specifically, the control tasks are broken down into individual regions and users, and the number of users in each region that need to participate in the control is determined, as well as the specific control values ​​for each user's equipment, including the power reduction ratio and output adjustment amount. Photovoltaic reverse overload warnings are implemented. Throughout the control process, power output is monitored in real time, and potential reverse overload situations are warned through analysis of power-related data. Control instructions are sent to the equipment end of each distributed photovoltaic user through a communication system that includes an electricity information collection system, a collection terminal, and a protocol converter.

[0017] Furthermore, in the formulation of photovoltaic control strategies, historical data and real-time monitoring data are used to dynamically evaluate the output capacity of photovoltaic power stations. The evaluation formula is: P out =f(E,T,D,M), where P out is the output capacity of the photovoltaic power station, E is the environmental factor vector, T is the equipment state vector, D is the power station design factor vector, M is the maintenance factor vector, and f is the functional relationship constructed by the machine learning algorithm.

[0018] Furthermore, in the formulation of photovoltaic control strategies, the photovoltaic power output in the future period is predicted based on the numerical meteorological forecast data and the trained power prediction model. The prediction formula is: P t =g(W t ,P t -1,P t -2,...), where P t is the photovoltaic power prediction value at time t, W t is the numerical weather forecast data at time t, P t-1 ,P t-2 ,... are the actual values ​​of photovoltaic power at the previous moment, the previous two moments,..., and g is the functional relationship constructed by the power prediction model.

[0019] Furthermore, in the implementation of the group control scheme, during the control process, the operating status and power output of the photovoltaic power station are monitored in real time, and the control strategy is dynamically adjusted according to the actual situation to ensure that the control effect meets the grid operation requirements. The calculation formula for adjusting the control in proportion to the user's rated capacity is as follows: Its P adj The middle is the power that the user needs to adjust, P user is the user's current actual power, R adj is the control index, R total It is the sum of the rated installed capacity of all users.

[0020] Furthermore, in the implementation of the group control scheme, the operating status and power output of the photovoltaic power station are monitored in real time, and the control strategy is dynamically adjusted according to the actual situation. The adjustment is based on the power deviation rate, and the power deviation rate calculation formula is: Where δ is the power deviation rate, P actual is the actual power, P target is the target power. When the power deviation rate exceeds the set threshold, the control strategy is adjusted.

[0021] Furthermore, in the implementation of the group adjustment and control scheme, the electricity consumption information collection system sends the control command to the collection terminal, and the collection terminal and the protocol converter use wired or wireless communication to transmit data. The protocol converter converts the control signal into instructions that the inverter can recognize, thereby realizing the adjustment of the inverter output.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. Accurate prediction and scientific decision-making

[0024] Through multi-source data fusion preprocessing, we comprehensively consider the influencing factors, lay a solid foundation for prediction, and improve accuracy; by building and optimizing the fusion model and training based on historical data, we provide a reliable basis for the control strategy.

[0025] 2. Intelligent control to stabilize the power grid

[0026] Through comprehensive analysis of multiple factors, the output capacity is evaluated in real time to provide precise guidance for regulation. By combining prediction with demand, precise regulation is carried out, and the communication system ensures the issuance of instructions, achieving group regulation and control, stabilizing grid operation, and balancing power during special periods.

[0027] 3. Clean, efficient and sustainable development

[0028] Solve the problem of photovoltaic grid access, give full play to the advantages of clean energy, and promote the transformation of energy structure; optimize operation, improve energy efficiency, reduce waste and emissions, achieve win-win benefits in multiple aspects, and promote sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a flow chart of the method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model of the present invention;

[0030] Figure 2 is a flow chart of photovoltaic power prediction model training and optimization of the present invention;

[0031] Figure 3 This is a flow chart of the photovoltaic group control strategy execution and dynamic adjustment of the present invention. DETAILED DESCRIPTION

[0032] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0033] like Figures 1 to 3 As shown, an embodiment of the present invention provides a method for predicting and group-adjusting low-voltage distributed photovoltaic output based on a big data model, including the following steps:

[0034] Step 1: Data collection and preprocessing

[0035] 1. Establish data connection interfaces with the distributed photovoltaic power station monitoring system, meteorological departments, and geographic information systems to ensure stable and real-time data transmission. For example, establish a dedicated line connection with the local weather station to obtain the latest meteorological data every 30 minutes, including information such as solar radiation intensity, temperature, humidity, air pressure, wind speed, and wind direction. At the same time, communicate with the distributed photovoltaic power station monitoring system via an Ethernet interface to collect historical output data, voltage, current, and other operating parameters of the photovoltaic power station in real time, and obtain geographic information data such as the latitude and longitude, altitude, and topography of the power station from the geographic information system.

[0036] 2. During the data cleaning process, a power fluctuation threshold of ±10% of rated power was set. When collected power data exceeded this threshold, it was identified as an abnormal data point and removed. For missing solar radiation intensity data, linear interpolation was used if the data changes between adjacent time points were relatively stable. If the data fluctuated significantly, mean interpolation was used, taking the average of other normal data points within the same time period on that day as the missing value.

[0037] 3. During data normalization, the maximum and minimum normalization method is used to map various data types to the [0, 1] interval. For example, for solar radiation intensity data, normalization allows data of different dimensions to be analyzed and processed on the same scale, providing a good data foundation for subsequent model training. In the field of data processing and analysis, especially in machine learning and statistical analysis, normalization to the [0, 1] interval is a common data preprocessing operation. For professionals working in this field, this is a basic concept and a common technical method, and they have a certain understanding of its principles and functions. In this sense, it can be considered common knowledge for those in the technical field.

[0038] Step 2: PV power prediction model training and optimization

[0039] 1. During the model research phase, we conducted a detailed analysis of the characteristics of traditional time series models (such as ARIMA, STL, and MLP), photovoltaic power generation physical models (such as PVLIB, component models, and inverter models), photovoltaic power generation transformer-based models (such as TFT, Autoformer, Informer, and Flowformer), and multi-task learning models. Taking a distributed photovoltaic power station with an installed capacity of 100kW as an example, we found that its output data exhibited a certain degree of seasonality and periodicity in time, and was closely related to meteorological factors such as solar radiation intensity and temperature. After comparative testing, we chose to use the photovoltaic power generation fusion physical model as the base model, introducing transformer and linear neural network models as supplements to construct a hybrid model.

[0040] 2. When constructing the integrated physical model, a multi-parameter physical model is established based on the equivalent circuit model of the photovoltaic cell, taking into account physical parameters such as the module's series resistance, parallel resistance, and diode characteristics. This model is combined with actual measured information such as the power station's inclination angle, azimuth, and module area. Through time series analysis of historical data, the weights of the impact of parameters such as irradiance and temperature on power output under different seasons and weather conditions are estimated, and the parameters in the physical model are optimized. For example, during the high temperatures of summer, temperature has a greater impact on power output. Time series analysis can be used to determine the correction coefficient for the temperature parameter, improving the model's prediction accuracy during this period.

[0041] 3. During model training, the power plant's historical data collected over the past year (after preprocessing) was divided into a training set, a validation set, and a test set, with the data split at 70%, 20%, and 10% respectively. The fusion model was trained using the training set, and the stochastic gradient descent algorithm was used to optimize model parameters. The learning rate was initially set to 0.01, and the regularization coefficient was 0.001. After each training cycle, model performance was evaluated using the validation set, with the root mean square error (RMSE) and mean absolute error (MAE) calculated to measure the deviation between the model's predicted and actual values. Based on the validation results, model hyperparameters were adjusted using cross-validation techniques. If signs of overfitting were detected, the regularization coefficient was increased or the learning rate was decreased as appropriate. After multiple training iterations, when the model's performance on the validation set stabilized and met the preset accuracy requirements (e.g., RMSE less than 5% of rated power), a final evaluation was performed using the test set. If the performance on the test set also met the requirements, model training was completed; otherwise, hyperparameters were adjusted and the model was retrained. New historical data is acquired regularly (e.g. monthly), added to the training set, and the model is retrained to adapt to changes in the power plant's operating status and environmental conditions.

[0042] Step 3: PV regulation strategy formulation

[0043] 1. Utilizing three years of historical and real-time monitoring data from the distributed photovoltaic power station, a neural network algorithm was used to establish a relationship model between photovoltaic output capacity and environmental factors (such as different weather types, such as sunny, cloudy, and rainy days, as well as different seasons and time periods), equipment status (such as determining module aging by monitoring module output current and voltage changes, and analyzing inverter efficiency based on the inverter's efficiency curve), plant design factors (such as a 30-degree module tilt angle, a south azimuth, and monocrystalline silicon module material), and maintenance factors (regular monthly maintenance). For example, on a cloudy, cool winter morning, the model calculated that the photovoltaic power station's output capacity was approximately 30% of its rated power.

[0044] 2. Based on numerical weather forecast data provided by the meteorological department (forecasted hourly solar radiation intensity and temperature for the next 72 hours) and the trained power prediction model, predict the photovoltaic power output for the future period. For example, between 9:00 AM and 10:00 AM, the solar radiation intensity is predicted to gradually increase, and the temperature will slowly rise. Combined with the model calculation, the photovoltaic power output during this period will gradually increase from 50 kW to 70 kW. Combined with the grid load demand forecast (80 kW for this period), the analysis shows that the photovoltaic power station needs to provide 10 kW of control power (80 kW - 70 kW) during this period, and the control time requirement is to be between 9:00 AM and 10:00 AM in real time based on power changes.

[0045] 3. According to the distributed photovoltaic distribution in the area where the photovoltaic power station is located (there are 50 distributed photovoltaic users in the area), user types (residential users, commercial users, etc.) and equipment performance (inverter maximum power tracking range, efficiency, etc.), the control tasks are decomposed to each user. For example, it is determined that the number of users who need to participate in the control in the area is 20, including 15 residential users and 5 commercial users. For residential users, the power reduction ratio of each residential user's equipment (inverter) is calculated based on their rated capacity ratio and control indicators. For example, if the rated capacity of residential user A is 5kW, the total rated installed capacity of all users participating in the control is 80kW, and the control indicator is to reduce the power by 10kW, then the power that residential user A needs to reduce is The power reduction ratio is Through the electricity consumption information collection system, the control instructions are accurately sent to the equipment end of each distributed photovoltaic user, realizing group control operations of the photovoltaic power station.

[0046] Step 4: Implementation of group coordination and control plan

[0047] 1. Automatically generate a list of regulated users based on the regulation strategy's scope (e.g., distributed PV users within a specific area), regulation indicators (e.g., total power reduction of 20kW), and user profiles (including user ratings, industry, and other information). For example, in an area with 100 distributed PV users, 30 eligible users are selected based on the regulation requirements and added to the list: 5 industrial users, 10 commercial users, and 15 residential users.

[0048] 2. According to the proportion of users' rated capacity, combined with the ratio of control index to rated installed capacity, users are regulated downward in proportion. Taking commercial user B as an example, its rated capacity is 8kW, and the total rated installed capacity of all users is 120kW. The control index is to reduce the power by 20kW. Then the power that commercial user B needs to adjust is The control command is sent to the collection terminal through the electricity consumption information collection system, and the data is transmitted between the collection terminal and the protocol converter using wireless communication (such as 4G network) to ensure the timeliness and stability of data transmission.

[0049] 3. The protocol converter accurately converts the received control signal into instructions that the inverter can recognize, thereby adjusting the inverter's output. For example, when the protocol converter receives an instruction to reduce power by 1.33kW, it converts it into a power adjustment signal for the inverter, causing the inverter to reduce its output power. During the regulation process, the operating status and power output of the photovoltaic power station are monitored in real time, and the power deviation rate is calculated every 5 minutes. If the power deviation rate is calculated to be 8% (with a set threshold of 5%) at a certain moment, exceeding the preset threshold, the regulation strategy is adjusted in a timely manner based on the deviation. For example, the number of users participating in the regulation can be increased or decreased, or the regulation value for each user can be recalculated to ensure that the regulation effect always meets the grid operation requirements and achieves power balance across the entire network.

[0050] The present invention's method for predicting and controlling low-voltage distributed photovoltaic output based on a big data model can achieve the following technical effects:

[0051] 1. Multi-source data integration, comprehensive consideration of influencing factors: This includes historical output data, various meteorological data (such as solar radiation intensity, temperature, humidity, air pressure, wind speed, wind direction, etc.), and geographic information data (plant latitude and longitude, altitude, topography). This data comprehensively reflects the various conditions that affect photovoltaic output. For example, different meteorological conditions directly affect photovoltaic power generation efficiency, geographical factors affect light reception and temperature conditions, and historical data contains the changing patterns of photovoltaic output. This provides a rich and comprehensive information foundation for accurate prediction, avoiding prediction errors caused by missing or one-sided data.

[0052] Advanced model construction improves forecast accuracy: A hybrid model is constructed that integrates physical models with machine learning methods, leveraging the strong interpretability of physical models based on principles (relying on geographical, meteorological, and equipment information) and the excellent ability of machine learning models to process complex nonlinear data. Based on the integrated physical model of photovoltaic power generation, numerous physical parameters (angle, area, shading, irradiance, temperature, etc.) are considered, and neural network models such as transformers and linear models are combined to supplement and optimize the expertise of time series data processing. This hybrid model can deeply explore the inherent relationships in the data, accurately capture the changing trends of photovoltaic power, effectively cope with various complex operating conditions, and provide a highly reliable basis for the scientific formulation of subsequent control strategies.

[0053] 2. Accurate strategic planning and execution to ensure a dynamic balance between supply and demand

[0054] Real-time dynamic assessment of PV output capacity: Using machine learning algorithms, a multi-factor model is constructed to analyze PV output capacity and the environment (weather, time of day, season, etc.), equipment (module aging, inverter efficiency, etc.), plant design (module orientation, tilt angle, materials, etc.), and maintenance (equipment maintenance frequency, etc.). This model calculates precise PV output capacity indicators under different conditions in real time. Regardless of environmental changes or fluctuations in equipment status, the system accurately assesses the actual output level of the PV plant, providing key data for the precise formulation of control strategies, eliminating blind adjustments and ensuring precise control.

[0055] Achieve precise matching and regulation based on supply and demand forecasts: Based on numerical weather forecasts and trained power prediction models, combined with grid load demand forecasts, the system meticulously analyzes the power and time requirements for PV power plant regulation during different time periods. When electricity demand fluctuates between peaks and valleys, it can predict and develop targeted regulation plans in advance, determining when, where, and for which regions and users to implement regulation. This allows for dynamic and precise matching of PV output with grid load demand, maintaining stable grid operation, avoiding grid failures caused by supply and demand imbalances, and ensuring stable and reliable power supply quality.

[0056] Real-time monitoring and dynamic adjustment to optimize regulation effects and adaptability: During the entire regulation process, the operating status and power output of the photovoltaic power station are monitored in real time and at high frequency (e.g., once every 5 minutes), and the power deviation rate is calculated in a timely manner. Once the power deviation exceeds the set threshold, the regulation strategy adjustment mechanism is immediately triggered. By increasing or decreasing the number of users participating in regulation, recalculating user regulation values ​​(power reduction ratio, output adjustment amount, etc.), and other flexible means, deviations can be quickly corrected to ensure that the regulation effect is always closely aligned with the actual needs of grid operation. This dynamic adjustment mechanism greatly enhances the grid's ability to accept distributed photovoltaics, enabling them to better integrate into the grid operation system, effectively respond to various complex and changing working conditions, and ensure the safe and stable operation of the grid.

[0057] 3. Balancing grid supply and demand, enhancing system stability and reliability: This system effectively addresses a series of grid challenges, including reverse overload of distribution transformers, reverse power flow at 10kV, and high voltage at the end of distribution lines, caused by the widespread integration of low-voltage distributed photovoltaic systems. During peak hours and extreme weather, precise control of photovoltaic output ensures a balanced supply and demand across the entire grid, protects the grid from impacts caused by fluctuations in photovoltaic output, enhances grid operation stability and reliability, and provides a continuous and stable power supply guarantee for social production and daily life.

[0058] Optimize clean energy allocation and promote energy structure upgrades and sustainable development: Enable low-voltage distributed photovoltaics to play a positive role in grid peak regulation, rationally adjust output according to grid load changes, reduce energy waste, and improve clean energy utilization efficiency. Promote the widespread application and deep integration of clean energy in the grid, promote the optimization and upgrading of the energy structure towards a clean and low-carbon direction, assist in the implementation of environmental protection and sustainable development strategies, and contribute to the construction of a green and low-carbon energy system.

[0059] Feasibility analysis of the scalability and sustainable development of the technical solution of the present invention's low-voltage distributed photovoltaic output prediction and group adjustment and control method based on a big data model:

[0060] 1. Scalability

[0061] (1) Model level

[0062] Flexible and scalable architecture: The hybrid model architecture, integrating physics and machine learning, allows for the flexible incorporation of new physical factors or advanced neural network structures. For example, as research deepens or new technologies emerge, the model can be rapidly adapted to low-voltage distributed photovoltaic systems of varying scales and scenarios, ensuring accurate power forecasting.

[0063] Data-driven optimization: Models trained on extensive historical data and regularly updated automatically adapt to changes in plant operations and the environment. For example, when the system scales up or equipment is upgraded, new data is incorporated into the training set, prompting the model to adaptively adjust parameters, maintaining predictive capabilities and reducing expansion costs.

[0064] (2) System architecture level

[0065] Common data interfaces facilitate expansion: Data interfaces established with multiple systems are based on common standards, facilitating access to a wider range of data sources. Whether adding new weather stations or integrating data from different power stations, these interfaces can be expanded through the expanded interface, supporting the diversification of the system's geographic coverage and data sources.

[0066] Distributed management of control strategies: In group control and group control solutions, control strategies can be implemented in a layered and distributed manner. As scale expands, additional control centers can be added or user groups can be subdivided. For example, with the development of distributed photovoltaic systems in cities, regional sub-control centers can be established to collaborate and ensure efficient operation.

[0067] 2. Sustainable Development Optimization

[0068] (1) Energy and environment

[0069] Power prediction optimization improves energy efficiency: Continuously optimize the power prediction model to accurately match power generation with load demand, reduce curtailment, and improve energy utilization. The system automatically adjusts prediction parameters based on factors such as climate change and component aging, adapting to environmental changes and achieving long-term, efficient power generation.

[0070] Control strategies adapt to environmental changes: Control strategies take environmental factors into account and can be further optimized based on climate research and environmental monitoring data. For example, seasonal strategies can be developed to adjust output in advance in response to extreme weather conditions, ensuring system coordination with the environment, minimizing negative impacts, and promoting sustainable development.

[0071] (2) Technology and industry adaptability

[0072] Keeping pace with technological trends: This system involves a variety of advanced technologies, allowing for the integration of new technologies and upgrades as they evolve. For example, advancements in big data processing and communications technologies can be leveraged to enhance data mining depth and control command transmission efficiency, maintaining system advancement and meeting future energy industry demands.

[0073] Integration of industry standards and specifications: Optimize the system based on the standards and specifications currently in development within the distributed photovoltaic industry. For example, compliance with standards for data interfaces, power prediction accuracy, and control strategy security ensures compatibility and compliance, promoting healthy industry development and system interoperability.

[0074] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for predicting and group regulating low-voltage distributed photovoltaic output based on a big data model, characterized in that: The following steps are involved: Step 1: Data collection and analysis: Collect historical output data, meteorological data and geographic information data of low-voltage distributed photovoltaics, and pre-process the collected data, including cleaning and normalization operations; Step 2: Construction of photovoltaic power prediction model: Based on the physical principles of photovoltaic power generation, a photovoltaic power generation prediction fusion model integrating multiple machine learning regression methods is constructed by considering various influencing factors, and the hyperparameters of the photovoltaic power generation prediction fusion model are corrected by fitting historical data, including model investigation and selection, and model training process; Step 3: Formulate photovoltaic control strategy: Conduct photovoltaic output capacity analysis, power forecast and control demand analysis, and determine the photovoltaic control strategy. The control strategy includes building a dynamic simulation model of output capacity, using NWP data to predict power, analyzing control demand and time requirements, formulating a plan based on regional output capacity indicators, and conducting photovoltaic reverse overload warning at the same time; Step 4: Implementation of group regulation and control scheme: Generate a list of regulated users according to the PV regulation strategy, and reduce the regulation in proportion to the user's rated capacity. Send control commands to the collection terminal through the power consumption information collection system, and then to the protocol converter, and finally adjust the inverter output to achieve group regulation and control.

2. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model according to claim 1, characterized in that: The meteorological data include solar radiation intensity, temperature, humidity, air pressure, wind speed and wind direction; the geographic information data include the latitude and longitude, altitude and topography of the power station; the cleaning operation includes removing obviously erroneous data points, supplementing the missing data by linear interpolation or mean interpolation method, and the normalization operation includes mapping the data to the [0,1] interval, establishing a data interface with the distributed photovoltaic power station monitoring system, meteorological department and geographic information system, and acquiring real-time data. The data acquisition time interval is every hour or shorter.

3. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model according to claim 1, characterized in that: In the training and optimization of the photovoltaic power prediction fusion model, after comparing various types of models, a hybrid model that integrates the physical model and the machine learning method is selected. Model training includes data partitioning, training, verification and testing operations; in the model training process, cross-validation technology is used to select the optimal model hyperparameter combination, and the model is retrained regularly based on newly acquired historical data. At least one of support vector machines, neural networks, and decision trees is selected as the regression method in the fusion model; based on the physical principles of photovoltaic power generation, a multi-parameter physical model is established. The multi-parameters include angle, area, shading, irradiance and temperature. The parameters in the physical model are estimated and optimized through time series analysis methods to construct a photovoltaic power prediction model.

4. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model as claimed in claim 3 is characterized in that: In the training and optimization of the photovoltaic power prediction model, the neural network models introduced are transformer and linear models. The bagging ensemble learning method is used to integrate multiple base models. The base models are trained by different sub-training sets obtained by random sampling with replacement of the training set.

5. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model according to claim 1, characterized in that: In the formulation of photovoltaic control strategies, machine learning algorithms are used to establish a relationship model between photovoltaic output capacity and environmental factors and equipment status, and photovoltaic output capacity indicators under different conditions are calculated in real time to build a dynamic simulation model of output capacity; NWP data is used to predict power, specifically, the photovoltaic power output in future time periods is predicted based on numerical meteorological forecast data and the trained photovoltaic power generation power prediction fusion model; the control demand and time requirements are analyzed, specifically, the control power and control time requirements that photovoltaic power stations need to provide in different time periods are analyzed in combination with the load demand forecast of the power grid; Formulate a plan based on regional output capacity indicators, specifically decompose the control tasks into various regions and users, determine the number of users in each region that need to participate in the control, and the specific control values ​​of each user's equipment, which include power reduction ratio and output adjustment; conduct photovoltaic reverse overload warning, that is, monitor power output in real time during the entire control process, and issue warnings for possible reverse overload situations through analysis of power-related data; send control instructions to the equipment end of each distributed photovoltaic user through the communication system, which includes a power consumption information collection system, a collection terminal and a protocol converter.

6. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model as claimed in claim 5, characterized in that: In the formulation of photovoltaic control strategy, historical data and real-time monitoring data are used to dynamically evaluate the output capacity of photovoltaic power stations. The evaluation formula is: P out =f(E,T,D,M), where P out is the output capacity of the photovoltaic power station, E is the environmental factor vector, T is the equipment state vector, D is the power station design factor vector, M is the maintenance factor vector, and f is the functional relationship constructed by the machine learning algorithm.

7. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model as claimed in claim 6, characterized in that: In the formulation of photovoltaic control strategy, the photovoltaic power output in the future period is predicted based on the numerical meteorological forecast data and the trained power prediction model. The prediction formula is: P t =g(W t ,P t -1,P t -2,...), where P t is the predicted photovoltaic power value at time t, W t is the numerical weather forecast data at time t, P t-1 ,P t-2 ,... are the actual values ​​of photovoltaic power at the previous moment, the previous two moments,..., and g is the functional relationship constructed by the power prediction model.

8. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model according to claim 1, characterized in that: In the implementation of the group control scheme, during the control process, the operating status and power output of the photovoltaic power station are monitored in real time, and the control strategy is dynamically adjusted according to the actual situation to ensure that the control effect meets the grid operation requirements. The calculation formula for adjusting the control in proportion to the user's rated capacity is as follows: Its P adj The power that the user needs to adjust, P user is the current actual power of the user, R adj is the control index, R total It is the sum of the rated installed capacity of all users.

9. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model as claimed in claim 8, characterized in that: In the implementation of the group control scheme, the operating status and power output of the photovoltaic power station are monitored in real time, and the control strategy is dynamically adjusted according to the actual situation. The basis for adjustment is the power deviation rate. The power deviation rate calculation formula is: Where δ is the power deviation rate, P actual is the actual power, P target is the target power. When the power deviation rate exceeds the set threshold, the control strategy is adjusted.

10. The method for predicting and group regulating and controlling low-voltage distributed photovoltaic output based on a big data model according to claim 9, characterized in that: During the implementation of the group control scheme, the electricity consumption information collection system sends control commands to the collection terminal. The collection terminal and the protocol converter use wired or wireless communication to transmit data. The protocol converter converts the control signal into instructions that the inverter can recognize, thereby realizing the adjustment of the inverter output.

Images