An ems control management system based on power prediction

Abstract

The application provides an EMS control management system based on power prediction, which is composed of a data acquisition and communication module, a power prediction module, an optimal scheduling module, an execution control module and a visualization and digital twin module; the system collects weather forecast data, power grid operation data and scheduling instruction data through the data acquisition and communication module, inputs the data after preprocessing into the power prediction module for multi-time scale power prediction, then inputs the predicted results into the optimal scheduling module for optimal scheduling instruction, inputs the optimal scheduling instruction into the execution control module for controlling the new energy power station, the energy storage system and the power grid connection point, and the data acquisition and communication module monitors the data of the new energy power station, the energy storage system and the power grid connection point in real time, and timely optimizes and adjusts; the visualization and digital twin module displays the data of each stage of system operation in real time, and the system provides strong technical support for operation and maintenance management and continuous optimization.

Description

Technical Field

[0001] This invention relates to the field of power management system technology, specifically to an EMS control and management system based on power prediction. Background Technology

[0002] With the rapid development of new energy power generation, the proportion of distributed power sources such as photovoltaic and wind power connected to the power system continues to increase, resulting in significant fluctuations and uncertainties in their power output. To ensure the safe, stable, and economical operation of the power grid, it is urgent to achieve accurate prediction of new energy output and intelligent scheduling of energy storage systems. The Energy Management System (EMS) plays a core role in this process, collecting power data, executing prediction algorithms, optimizing operating strategies, and issuing control commands. Currently, traditional EMS systems are mostly based on static monitoring and manual strategies, lacking data-driven dynamic prediction capabilities and anomaly self-healing mechanisms, making it difficult to cope with power fluctuations and multi-objective optimization requirements brought about by new energy integration. At the same time, traditional systems have limited capabilities in visualization and model mapping, failing to achieve simulation and digital twin display of actual operating states. Therefore, based on the needs of power system management, the applicant proposes an EMS control and management system based on power prediction. This system is an EMS control and management system that integrates power prediction, intelligent scheduling, anomaly detection and self-healing control, and digital twin visualization technologies. It can achieve proactive optimization and adaptive control of the system, significantly improving the utilization rate of new energy and operational reliability, and possesses significant engineering application value. Summary of the Invention

[0003] To address the aforementioned technical challenges, this invention proposes an EMS control and management system based on power prediction. The system comprises a data acquisition and communication module, a power prediction module, an optimization scheduling module, an execution control module, and a visualization and digital twin module. By introducing anomaly detection and self-healing control functions, the system can automatically identify communication interruptions, power deviations, and equipment failures and execute safety recovery strategies to ensure continuous and stable system operation. Through visualization and digital twin technology, the system achieves synchronous display of the physical system and the virtual model, supporting operation monitoring, historical playback, and model optimization. This system significantly improves prediction accuracy, scheduling efficiency, operational reliability, and visualization level, making it suitable for photovoltaic-storage microgrids and distributed energy scenarios.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] An EMS control and management system based on power prediction is characterized by comprising a data acquisition and communication module, a power prediction module, an optimization scheduling module, an execution control module, and a visualization and digital twin module. The data acquisition and communication module is responsible for acquiring power grid operation data from multiple internal and external channels and issuing control commands. The power prediction module performs power prediction based on the received data using a combination of physical and statistical learning methods. The optimization scheduling module solves for the optimal energy storage charging and discharging strategy based on the probabilistic prediction results. The execution control module is responsible for converting the optimization commands into safe and reliable actions. The visualization and digital twin module constructs a virtual simulation model based on digital twin technology, realizing a bidirectional mapping between the physical and virtual systems for visualization.

[0006] Furthermore, the data acquisition and communication module of the power prediction-based EMS control and management system is responsible for acquiring power grid operation data from multiple internal and external channels and issuing control commands as follows:

[0007] External data acquisition: High-precision numerical weather forecast data, including irradiance sequences for specific future time periods, are obtained from professional meteorological service agencies via API interfaces. Wind speed sequence and ambient temperature sequence ;

[0008] Internal real-time and historical data acquisition: Real-time meteorological data of new energy power plants are collected based on the power plant monitoring system. , , Historical power output data Real-time load data State of charge of energy storage system and power grid dispatch instructions ;

[0009] Control command data output: The preprocessed data is output to the power prediction module; the subsequent communication module will optimize the control commands generated by the scheduling module. The command is sent to the execution control module, which then controls and adjusts the energy storage converter and other execution units according to the command.

[0010] Furthermore, the prediction model of the power prediction module in the power prediction-based EMS control and management system employs deep learning models such as Long Short Memory Network (LSTM) and Temporal Convolutional Network (TCN), with its core being the learning of a mapping function. The formula is:

[0011]

[0012] in: yes Predicted power at time;

[0013] To predict the step size;

[0014] Input feature vector include: .

[0015] Furthermore, the multi-timescale and probabilistic prediction of the power prediction module in the power prediction-based EMS control and management system specifically includes:

[0016] The current forecast predicts power output for the next 24-72 hours at 15-minute intervals, which will be used to develop a preliminary scheduling plan.

[0017] Intraday rolling forecast: The forecast for the next 4-6 hours is updated every 5-15 minutes to adjust the plan.

[0018] Real-time prediction: Predicts power output for the next 15-30 minutes at 1-5 minute intervals for real-time control;

[0019] Probability prediction: Output the quantile interval of the predicted value ,in To determine the confidence level, or to generate N equally probable future output scenarios using scene generation techniques. To quantify uncertainty.

[0020] Furthermore, the objective function of the optimization scheduling module in the power prediction-based EMS control and management system aims to minimize total operating cost or maximize revenue, and its mathematical expression is:

[0021]

[0022] in: Let t be the grid electricity price at time t, where a positive value represents the cost of purchasing electricity and a negative value represents the revenue from selling electricity.

[0023] It is the power exchanged with the power grid. = - - ;

[0024] It is the energy storage loss cost coefficient;

[0025] It represents the charging and discharging power of energy storage; charging is positive and discharging is negative.

[0026] Robust / stochastic optimization considering uncertainty: Conditional value at risk or robust optimization methods are used to handle prediction errors.

[0027] Furthermore, the execution control module of the power prediction-based EMS control management system is responsible for translating optimization instructions into safe and reliable actions, specifically as follows:

[0028] Command translation: Receives power commands from the optimized scheduling module. This is converted into a control signal that the energy storage converter can recognize;

[0029] Closed-loop correction: Introducing feedback control based on real-time power deviation, the actual controlled power... ;

[0030] =

[0031] in: = This refers to the real-time power prediction error.

[0032] This is the proportionality coefficient;

[0033] The integral coefficient;

[0034] Anomaly Detection and Self-Healing Control: An anomaly detection and self-healing mechanism is introduced to perform real-time analysis of equipment status, current and voltage waveforms, power deviations, and communication delays. Rule-based diagnostics and machine learning algorithms are used to identify anomalies. When communication interruptions, sensor drift, or power anomalies are detected, the system automatically executes a self-healing strategy.

[0035] Communication error: Switch to local cached data and maintain safe power output;

[0036] Power anomaly: Implement limiting and soft handover strategies to prevent power oscillations;

[0037] Energy storage failure: Automatically isolates the faulty unit and reconstructs the control logic to ensure continuous operation.

[0038] Furthermore, the specific performance of the visualization and digital twin module in the power prediction-based EMS control and management system is as follows:

[0039] 1) Multi-dimensional visualization monitoring: Real-time display of key indicators such as photovoltaic output, energy storage SOC, electricity price and prediction deviation through power flow graphs, 3D scenes and trend curves;

[0040] 2) Runtime playback and simulation analysis: Supports the reproduction of historical operating conditions and parameter playback, helping operation and maintenance personnel to analyze prediction errors and optimization effects, and improve the interpretability of the model;

[0041] 3) Digital twin linkage: The system maps the operating status of equipment in real time in virtual space, performs predictive deviation analysis and control behavior verification, and realizes the "what you see is what you get" operation monitoring experience.

[0042] The benefits of this application are:

[0043] 1. The EMS control and management system based on power prediction introduces deep learning models such as Long Short-Term Memory (LSTM) networks or Temporal Convolutional Networks (TCNs), combined with meteorological data, historical power output, and load information, to achieve power prediction at multiple time scales (day-ahead, intraday, and real-time). This method can significantly improve prediction accuracy, effectively reduce the uncertainty of renewable energy output, and save approximately 15%-20% of daily electricity costs through precise "low-storage, high-generation" strategies.

[0044] 2. The EMS control and management system based on power prediction effectively smooths out fluctuations in photovoltaic power, significantly reduces the impact on the upstream power grid, and has good stability.

[0045] 3. The EMS control and management system based on power prediction has the advantage of low power consumption, is suitable for passive wireless sensors, and increases battery life.

[0046] 4. The EMS control and management system based on power prediction adopts an energy storage dispatch algorithm that combines robust optimization and stochastic optimization. Under the premise of ensuring grid stability, it takes into account the cost of electricity purchase, energy storage life and the utilization rate of new energy sources, so as to achieve optimal control of energy storage charging and discharging, and improve the system economy and operating efficiency.

[0047] 5. The EMS control and management system based on power prediction introduces anomaly detection and self-healing mechanisms. The system can automatically enter a safe mode in the event of communication failure, power deviation or equipment abnormality to ensure continuous operation.

[0048] 6. The EMS control and management system based on power prediction achieves real-time display and historical playback of operating status through virtual simulation and multi-dimensional visualization, greatly improving the system's transparency and operability. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the system architecture of the present invention;

[0050] Figure 2 This is a schematic diagram of the power prediction module's workflow in this invention. Detailed Implementation

[0051] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0052] like Figure 1-2As shown, this is an EMS control and management system based on power prediction. The system comprises a data acquisition and communication module, a power prediction module, an optimization scheduling module, an execution control module, and a visualization and digital twin module. The data acquisition and communication module is responsible for acquiring grid operation data from multiple internal and external channels and issuing control commands. The power prediction module uses a combination of physical and statistical learning methods to predict power based on the received data. The optimization scheduling module solves for the optimal energy storage charging and discharging strategy based on the probabilistic prediction results. The execution control module is responsible for translating the optimization commands into safe and reliable actions. The visualization and digital twin module constructs a virtual simulation model based on digital twin technology, realizing a two-way mapping between the physical and virtual systems for visualization.

[0053] The data acquisition and communication module of the power prediction-based EMS control and management system shown is responsible for acquiring power grid operation data from multiple internal and external channels and issuing control commands as follows:

[0054] External data acquisition: High-precision numerical weather forecast data, including irradiance sequences for specific future time periods, are obtained from professional meteorological service agencies via API interfaces. Wind speed sequence and ambient temperature sequence ;

[0055] Internal real-time and historical data acquisition: Real-time meteorological data of new energy power plants are collected based on the power plant monitoring system. , , Historical power output data Real-time load data State of charge of energy storage system and power grid dispatch instructions ;

[0056] Control command data output: The preprocessed data is output to the power prediction module; the subsequent communication module will optimize the control commands generated by the scheduling module. The command is sent to the execution control module, which then controls and adjusts the energy storage converter and other execution units according to the command.

[0057] The power prediction module of the EMS control and management system shown employs deep learning models such as Long Short Memory Network (LSTM) and Temporal Convolutional Network (TCN), with its core function being the learning of a mapping function. The formula is:

[0058]

[0059] in: yes Predicted power at time;

[0060] To predict the step size;

[0061] Input feature vector include: .

[0062] The multi-timescale and probabilistic prediction of the power prediction module in the power prediction-based EMS control and management system shown is as follows:

[0063] The current forecast predicts power output for the next 24-72 hours at 15-minute intervals, which will be used to develop a preliminary scheduling plan.

[0064] Intraday rolling forecast: The forecast for the next 4-6 hours is updated every 5-15 minutes to adjust the plan.

[0065] Real-time prediction: Predicts power output for the next 15-30 minutes at 1-5 minute intervals for real-time control;

[0066] Probability prediction: Output the quantile interval of the predicted value ,in To determine the confidence level, or to generate N equally probable future output scenarios using scene generation techniques. To quantify uncertainty.

[0067] The objective function of the optimization scheduling module in the power prediction-based EMS control and management system shown is to minimize total operating cost or maximize revenue. Its mathematical expression is:

[0068]

[0069] in: Let t be the grid electricity price at time t, where a positive value represents the cost of purchasing electricity and a negative value represents the revenue from selling electricity.

[0070] It is the power exchanged with the power grid. = - - ;

[0071] It is the energy storage loss cost coefficient;

[0072] It represents the charging and discharging power of energy storage; charging is positive and discharging is negative.

[0073] Robust / Stochastic Optimization Considering Uncertainty: To handle prediction errors, conditional value at risk or robust optimization methods are employed; for example, in stochastic optimization, the objective function becomes minimizing the expected cost.

[0074]

[0075] in: It is the probability of the i-th predicted scenario;

[0076] Power balance: = ;

[0077] Energy Storage SOC Dynamics:

[0078] ( <0)

[0079] ( >0)

[0080] SOC upper and lower limits: ;

[0081] Charge and discharge power limits: .

[0082] The execution control module of the power prediction-based EMS control management system shown is responsible for translating optimization commands into safe and reliable actions. Specifically:

[0083] Command translation: Receives power commands from the optimized scheduling module. This is converted into a control signal that the energy storage converter can recognize;

[0084] Closed-loop correction: Introducing feedback control based on real-time power deviation, the actual controlled power... ;

[0085] =

[0086] in: = This refers to the real-time power prediction error.

[0087] This is the proportionality coefficient;

[0088] The integral coefficient;

[0089] Anomaly Detection and Self-Healing Control: An anomaly detection and self-healing mechanism is introduced to perform real-time analysis of equipment status, current and voltage waveforms, power deviations, and communication delays. Rule-based diagnostics and machine learning algorithms are used to identify anomalies. When communication interruptions, sensor drift, or power anomalies are detected, the system automatically executes a self-healing strategy.

[0090] Communication error: Switch to local cached data and maintain safe power output;

[0091] Power anomaly: Implement limiting and soft handover strategies to prevent power oscillations;

[0092] Energy storage failure: Automatically isolates the faulty unit and reconstructs the control logic to ensure continuous operation.

[0093] The specific performance of the visualization and digital twin module in the power prediction-based EMS control and management system shown is as follows:

[0094] 1) Multi-dimensional visualization monitoring: Real-time display of key indicators such as photovoltaic output, energy storage SOC, electricity price and prediction deviation through power flow graphs, 3D scenes and trend curves;

[0095] 2) Runtime playback and simulation analysis: Supports the reproduction of historical operating conditions and parameter playback, helping operation and maintenance personnel to analyze prediction errors and optimization effects, and improve the interpretability of the model;

[0096] 3) Digital twin linkage: The system maps the operating status of equipment in real time in virtual space, performs predictive deviation analysis and control behavior verification, and realizes the "what you see is what you get" operation monitoring experience.

[0097] The EMS control and management system based on power prediction shown below is an example of an industrial park microgrid that includes a photovoltaic power station, an energy storage system, and local loads:

[0098] 1) System hardware and software configuration;

[0099] Hardware platform: The system is deployed on an industrial server with sufficient computing power and storage space, and communicates with the monitoring system, weather station, energy storage converter and power grid dispatch center in the site via industrial Ethernet;

[0100] Software environment: The operating system is Linux; the core algorithm modules (power prediction and optimized scheduling) are written in C++ and integrated into a unified energy management software platform; the database is a time-series database for efficient storage and querying of massive amounts of real-time data.

[0101] External interfaces: Real-time data from the SCADA system of the park are obtained from the photovoltaic inverter, energy storage PCS and electricity meter via OPC UA or IEC 104 protocol; numerical weather forecasts for the next 7 days are obtained from commercial meteorological service providers via HTTP / HTTPS API, with an update frequency of once per hour; charging and discharging power commands are sent to the energy storage converter via Modbus TCP protocol.

[0102] 2) Implementation scenarios and parameters;

[0103] Scenario: An industrial park in East China has a 5MW photovoltaic power station and a 2MW / 4MWh lithium iron phosphate battery energy storage system. The typical daily load of the park fluctuates between 3MW and 8MW; the power grid implements peak-valley pricing.

[0104] Key parameters:

[0105] Rated energy storage capacity: ;

[0106] Maximum charging / discharging power of energy storage: ;

[0107] Energy storage SOC operating range: ;

[0108] Energy storage charge / discharge efficiency: ;

[0109] 3) Detailed description of the system workflow;

[0110] Taking a certain day in summer as an example, the system operates according to the following process:

[0111] Data collection began the previous day;

[0112] The data acquisition and communication module began acquiring 24-hour weather forecast data from the meteorological API at 14:00 that day. , , And obtain historical 72-hour photovoltaic output and load data from the SCADA system;

[0113] The forecast and optimization were made the day before and will be implemented at 14:30 the previous day.

[0114] The power prediction module is triggered, running the day-ahead prediction model. The model input includes the aforementioned weather forecast and historical data, and the output is a sequence of photovoltaic power output predictions at 15-minute intervals from 00:00 to 24:00 the following day. and its 90% confidence interval ;

[0115] The optimization scheduling module is based on the point prediction sequence and known peak and valley electricity prices, such as 1.2 yuan / kWh during peak hours, 0.7 yuan / kWh during normal hours, and 0.3 yuan / kWh during valley hours. The module performs optimization calculations with the objective function of "minimizing electricity purchase cost".

[0116] Optimization results: Generate the energy storage charging and discharging plan for the next day; for example, the plan is to charge at maximum power during off-peak hours (00:00-08:00) to increase the SOC from 0.2 to 0.95; and to discharge during peak hours (18:00-22:00) to reduce the peak power purchased by the park from the grid;

[0117] Daily rolling corrections, made in real time on the same day;

[0118] Every 15 minutes, the system initiates a rolling intraday forecast, using the latest real-time meteorological data and ultra-short-term weather forecasts to update the photovoltaic output forecast for the next 4 hours. ;

[0119] Based on this updated and more accurate forecast, as well as the current actual SOC of the energy storage, the optimized scheduling module makes rolling adjustments to future charging and discharging plans.

[0120] Scenario Example: At 10:00 AM, rolling forecasts indicated that afternoon cloud cover would be higher than previously predicted, leading to a 1MW reduction in photovoltaic output. The optimization module then adjusted its approach: reducing planned charging power during normal hours (11:00 AM - 2:00 PM) to reserve more power for the evening peak demand, thus avoiding the risk of having to purchase electricity at higher prices during peak hours due to insufficient photovoltaic output.

[0121] Real-time control and closed-loop calibration are performed on the same day.

[0122] Every minute, the system initiates real-time forecasting to predict the photovoltaic output for the next 15 minutes. ;

[0123] The execution control module receives the current minute-level power command issued by the optimization scheduling module. For example, -500kW means a discharge of 500kW;

[0124] Meanwhile, the module monitors the actual output of the photovoltaic system in real time. Suppose that due to a cloud passing by, the actual power output is 100kW lower than the real-time forecast, i.e., the error... ;

[0125] The execution control module immediately uses its PI controller for calibration:

[0126] =

[0127] It adjusts the actual discharge power from -500kW to -450kW, discharging an additional 50kW to partially compensate for the power deficit caused by clouds and maintain the stability of power exchange with the grid; this is a typical "feedforward + feedback" control process.

[0128] 4) Visualization and recording throughout the entire process;

[0129] Throughout the process, operators can clearly see the following on the visual interface:

[0130] Comparison of the three curves—the day-ahead forecast, the intraday forecast, and the real-time forecast—with the actual photovoltaic output curve;

[0131] Planned curves and actual changes in energy storage SOC;

[0132] The system introduces a digital twin model to map the operating status of photovoltaic arrays, energy storage devices, and load nodes in a three-dimensional interface in real time, enabling visualization of energy flow, equipment status, and the execution of scheduling commands.

[0133] Meanwhile, the anomaly detection module monitors the equipment's operating status, power deviation, and communication links in real time. If there is a limit violation, lag, or fault, the interface will automatically generate an alarm and link the self-healing control strategy to achieve automatic power limiting, switching to backup paths, or restarting to ensure stable system operation.

[0134] All operational data, prediction records, control commands, alarm events, and operation logs are saved in time series to provide complete data support for subsequent performance analysis, model training, and operational auditing.

[0135] 5) Real-time economic indicators of the system, such as the cumulative electricity savings for the day;

[0136] All data, including the predicted value, commanded value, actual value, and error at each moment, are recorded for subsequent performance analysis and model retraining.

[0137] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any other way. Any modifications or equivalent changes made based on the technical essence of the present invention shall still fall within the scope of protection claimed by the present invention.

Claims

1. A power prediction based EMS control management system, characterized by: The power prediction-based EMS control and management system comprises a data acquisition and communication module, a power prediction module, an optimization scheduling module, an execution control module, and a visualization and digital twin module. The data acquisition and communication module is responsible for acquiring grid operation data from multiple internal and external channels and issuing control commands. The power prediction module uses a combination of physical and statistical learning methods to predict power based on the received data. The optimization scheduling module solves for the optimal energy storage charging and discharging strategy based on the probabilistic prediction results. The execution control module is responsible for translating the optimization commands into safe and reliable actions. The visualization and digital twin module constructs a virtual simulation model based on digital twin technology, achieving bidirectional mapping between the physical and virtual systems for visualization.

2. The power prediction based EMS control management system of claim 1, wherein: The data acquisition and communication module of the power prediction-based EMS control and management system is responsible for acquiring power grid operation data from multiple internal and external channels and issuing control commands. External data acquisition: High-precision numerical weather forecast data, including irradiance sequences for specific future time periods, are obtained from professional meteorological service agencies via API interfaces. Wind speed sequence and ambient temperature sequence ; Internal real-time and historical data acquisition: Real-time meteorological data of new energy power plants are collected based on the power plant monitoring system. , , Historical power output data Real-time load data State of charge of energy storage system and power grid dispatch instructions ; Control command data output: The preprocessed data is output to the power prediction module; the subsequent communication module will optimize the control commands generated by the scheduling module. The command is sent to the execution control module, which then controls and adjusts the energy storage converter and other execution units according to the command.

3. The power prediction based EMS control management system of claim 1, wherein: The prediction model of the power prediction module of the power prediction-based EMS control management system is set by using a deep learning model such as a long short-term memory network model and a time sequence convolution network, and the core is to learn a mapping function , and the formula is: ; wherein: is predicted power at the instant of time; to predict a step size; input feature vector comprising: .

4. The power prediction based EMS control management system of claim 1, wherein: The power prediction module of the power prediction management system based on power prediction features multi-timescale and probabilistic prediction, specifically: The current forecast predicts power output over the next 24-72 hours at 15-minute intervals, which will be used to develop a preliminary scheduling plan. Intraday rolling forecast: The forecast for the next 4-6 hours is updated every 5-15 minutes to adjust the plan. Real-time prediction: Predicts power output for the next 15-30 minutes at 1-5 minute intervals for real-time control; Probability prediction: Output the quantile interval of the predicted value ,in To determine the confidence level, or to generate N equally probable future output scenarios using scene generation techniques. To quantify uncertainty.

5. The EMS control and management system based on power prediction according to claim 1, characterized in that: The objective function of the optimization scheduling module in the power prediction-based EMS control and management system aims to minimize total operating cost or maximize revenue. Its mathematical expression is as follows: ; wherein: is the grid electricity price at time t, positive for electricity purchase cost and negative for electricity sale revenue; is the exchanged power with the grid, ;​​​ is the loss cost coefficient of the energy storage; is the charge-discharge power of the energy storage, positive for charging and negative for discharging; Robust / stochastic optimization considering uncertainty: Conditional value at risk or robust optimization methods are used to handle prediction errors.

6. The EMS control and management system based on power prediction according to claim 1, characterized in that: The execution control module of the power prediction-based EMS control and management system is responsible for translating optimization instructions into safe and reliable actions. Specifically, it does the following: Instruction conversion: receive power instructions of the optimization scheduling module , and convert them into control signals recognizable by the energy storage converter Closed loop correction: Introduce feedback control based on real-time power deviation, actual control power ; = ; wherein: = is the real-time power prediction error; is a proportionality factor; is the integral coefficient; Anomaly Detection and Self-Healing Control: An anomaly detection and self-healing mechanism is introduced to perform real-time analysis of equipment status, current and voltage waveforms, power deviations, and communication delays. Rule-based diagnostics and machine learning algorithms are used to identify anomalies. When communication interruptions, sensor drift, or power anomalies are detected, the system automatically executes a self-healing strategy. Communication error: Switch to local cached data and maintain safe power output; Power anomaly: Implement limiting and soft handover strategies to prevent power oscillations; Energy storage failure: Automatically isolates the faulty unit and reconstructs the control logic to ensure continuous operation.

7. The EMS control and management system based on power prediction according to claim 1, characterized in that: The specific performance of the visualization and digital twin module in the power prediction-based EMS control and management system is as follows: 1) Multi-dimensional visualization monitoring: Real-time display of key indicators such as photovoltaic output, energy storage SOC, electricity price and prediction deviation through power flow graphs, 3D scenes and trend curves; 2) Runtime playback and simulation analysis: Supports the reproduction of historical operating conditions and parameter playback, helping operation and maintenance personnel to analyze prediction errors and optimization effects, and improve the interpretability of the model; 3) Digital twin linkage: The system maps the operating status of equipment in real time in virtual space, performs predictive deviation analysis and control behavior verification, and realizes the "what you see is what you get" operation monitoring experience.

Images