Power grid flexible control and optimal scheduling system for distributed energy access

Abstract

The invention relates to the technical field of intelligent scheduling, in particular to a distributed energy access-oriented power grid flexible control and optimal scheduling system, which comprises a dynamic sensing module, a prediction correction module, a flexible scheduling module, a self-adaptive control module and a collaborative traceability module. In the scheme, the dynamic sensing module of the system can capture fine fluctuation of distributed energy at high frequency, and second-level acquisition is started once the output change rate exceeds a certain range, which is far higher than the response speed of a traditional system. By means of the real-time and accurate sensing capacity, the system can rapidly capture energy at the initial stage of energy change, precious time is won for follow-up prediction and scheduling, it is ensured that the system closely follows the energy dynamic state all the time, the control force on distributed energy is greatly improved, the system can respond to various sudden changes in time, and the continuity of energy supply is effectively guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of intelligent dispatching technology, and in particular to a power grid flexible control and optimized dispatching system for distributed energy access. Background Art

[0002] Distributed energy sources are increasingly becoming a part of the power system. For example, many industrial plants are actively developing distributed photovoltaic power generation projects to achieve self-generation and self-consumption of electricity. At the same time, distribution networks, as a key cornerstone of the new power system, are shouldering the heavy responsibility of integrating large-scale renewable energy. With the widespread adoption of distributed energy, the volatility and intermittency of its output are challenging traditional grid dispatching and control methods, necessitating innovative systems to ensure the stability and reliability of power supply.

[0003] Currently, facing the large-scale integration of distributed energy resources, existing systems have shortcomings in real-time perception, accurate prediction, optimized scheduling, and adaptive control. Traditional perception methods struggle to capture rapid changes in energy output at high frequencies, resulting in prediction models unable to adapt promptly to actual conditions. This, in turn, causes scheduling instructions to become disconnected from actual energy fluctuations, and control strategies to be unable to effectively maintain grid frequency and voltage stability. When distributed energy output fluctuates abnormally, existing systems are unable to quickly adjust the acquisition frequency, making it difficult to obtain accurate data in a timely manner. This results in a lack of reliable basis for subsequent prediction, scheduling, and control, impacting the stable operation of the system. Therefore, we propose a flexible control and optimized scheduling system for the power grid that is designed for distributed energy integration. Summary of the Invention

[0004] In response to the deficiencies of the prior art, the present invention provides a flexible control and optimized dispatching system for a power grid for distributed energy access, thereby solving the technical problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0006] A flexible control and optimized dispatching system for distributed energy access, including a dynamic perception module, a prediction and correction module, a flexible dispatching module, an adaptive control module, and a collaborative traceability module;

[0007] The dynamic perception module includes a high-frequency acquisition unit, an anomaly location unit, and a data repair unit, which are used to capture the output fluctuations and anomalies of distributed energy in real time and trigger a hierarchical processing mechanism to provide data for subsequent prediction, scheduling, and control;

[0008] The prediction correction module includes a deviation analysis unit, a model update unit, and an instruction backtracking unit, which is used to dynamically correct the prediction model through timing deviation propagation analysis to ensure that the scheduling instructions are in line with the actual situation;

[0009] The flexible scheduling module includes a cost optimization unit, a voltage stabilization unit, and a real-time correction unit, which are used to generate scheduling instructions based on multi-objective optimization and dynamically respond to fluctuations in distributed energy resources;

[0010] The adaptive control module includes a VSG frequency regulation unit, an island switching unit, and a damping adjustment unit, which are used to maintain grid frequency and voltage stability through parameter adaptation and multi-level response strategies;

[0011] The collaborative tracing module includes an event storage unit, a root cause analysis unit, and a policy self-update unit, which are used to record the entire system operation chain and support fault backtracking and policy optimization.

[0012] In one possible implementation, the high-frequency acquisition unit of the dynamic sensing module starts second-level acquisition when the output change rate of distributed energy exceeds 15%, and if the change rate is less than 5% within three consecutive sampling periods, the default minute-level acquisition frequency is restored;

[0013] The abnormality location unit distinguishes between transient disturbance and equipment failure based on three time series data. t+1 ≥0.9P t And P t ≤0.5P r When, P r is the rated power of the equipment, P t If it is the current power, it is determined to be a device failure; otherwise, it is determined to be a transient disturbance;

[0014] The data repair unit uses a sliding average filtering method to repair instantaneous disturbances, and evaluates the repair effect by comparing the residual sum of squares of the data before and after repair. If the repair fails, the sliding window size is adjusted or a backup data source is started.

[0015] In a possible implementation, the deviation analysis unit of the prediction correction module is used to calculate the prediction deviation and determine whether it is a trend deviation. t >0.1P r and e t+1 >e t When, the prediction deviation e t =|P pr -P ac |,P pr is the predicted power, P ac is the actual power, P r The rated power of the equipment is determined to be a trend deviation and requires a global model update;

[0016] The model updating unit dynamically adjusts the meteorological weight of the LSTM model for single-point deviations and evaluates the update effect through the root mean square error of the cross-validation set. If the update effect is not good, the model parameters are rolled back and the hyperparameter optimization process is started;

[0017] When the deviation causes the scheduling instruction to fail, the instruction backtracking unit backtracks to the deviation starting point and regenerates the optimization instruction, and pushes the backtracking result to the adaptive control module.

[0018] In one possible implementation, the cost optimization unit of the flexible scheduling module optimizes the solution with the goal of minimizing the power purchase cost and network loss, and evaluates the optimization results through a feasibility verification matrix. If the solution is not feasible, constraints are automatically added and the solution is optimized again.

[0019] The voltage stabilization unit constrains node voltage deviations within a reasonable range and uses a three-level early warning mechanism to trigger corresponding operations of the dynamic perception module and the adaptive control module;

[0020] The real-time correction unit updates the scheduling plan every 5 minutes and evaluates the necessity of correction based on the accumulated prediction deviation. If the execution effect after correction is not good, the correction cycle is shortened and the prediction correction module is triggered to re-evaluate the accuracy of the prediction model.

[0021] In one possible implementation, the VSG frequency regulation unit of the adaptive control module simulates the inertia response of the synchronous machine through a control algorithm to maintain the stability of the grid frequency, and evaluates the regulation effect through the frequency change rate. When the frequency is abnormal, it feeds back to the flexible scheduling module to adjust the constraint conditions;

[0022] The island switching unit triggers island operation switching according to a multi-criteria fusion algorithm when a power grid fault occurs, and monitors the frequency and voltage in the island in real time through the PMU;

[0023] The damping adjustment unit dynamically adjusts the damping coefficient based on the Lyapunov criterion and evaluates the damping effect through system eigenvalue analysis. If oscillation still exists after the damping adjustment, the VSG frequency modulation unit is combined for coordinated control.

[0024] In one possible implementation, the event storage unit of the collaborative traceability module stores the "perception-scheduling-control" full-link data by timestamp and verifies the data integrity through a hash check mechanism;

[0025] The root cause analysis unit uses Bayesian network reasoning to locate the fault source when data is abnormal, and pushes the analysis results to the model update unit and the strategy self-update unit;

[0026] The strategy self-update unit automatically replaces the strategy with an alternative plan when the strategy fails, and evaluates the strategy failure through key performance indicators. The new strategy is synchronized to all modules after verification.

[0027] Beneficial effects compared with existing technologies:

[0028] 1. In this solution, the system's dynamic sensing module can capture subtle fluctuations in distributed energy at high frequency. Once the output change rate exceeds a certain range, it initiates acquisition within seconds, far exceeding the response speed of traditional systems. This real-time, precise sensing capability allows the system to quickly detect energy changes at their earliest stages, buying valuable time for subsequent prediction and scheduling. This ensures that the system always keeps pace with energy dynamics, significantly improving its control over distributed energy, enabling the system to promptly respond to various sudden changes and effectively ensuring the continuity of energy supply.

[0029] 2. In this solution, the prediction correction module uses unique time series deviation propagation analysis to accurately identify the type of prediction deviation and promptly adjust the prediction model weights, significantly improving prediction accuracy. The flexible scheduling module, based on multi-objective optimization, takes into account both electricity purchase costs and minimizing network losses to generate scientific scheduling instructions. The two modules work together to ensure that the scheduling plan closely matches the actual energy output, avoiding scheduling inaccuracies caused by prediction deviations in traditional systems, optimizing energy distribution, improving energy utilization efficiency, reducing operating costs, and enhancing the economic operation level of the system.

[0030] 3. This solution utilizes adaptive parameters and a multi-stage response strategy through the adaptive control module. For example, the VSG frequency modulation unit simulates the inertia of synchronous machines to maintain frequency stability, and the island switching unit rapidly switches to ensure power supply to critical loads in the event of a grid fault. These measures provide the system with strong stability and reliability, enabling stable operation even under complex operating conditions. Furthermore, the collaborative traceability module records all system operations, providing a basis for fault tracing and strategy optimization, further improving the system's control strategy and continuously enhancing the system's ability to cope with complex grid environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention with reference to the accompanying drawings.

[0032] Figure 1 Schematic diagram of the control and optimization scheduling system framework of the present invention;

[0033] Figure 2 It is a flow chart of the control and optimization scheduling system of the present invention. DETAILED DESCRIPTION

[0034] Preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the present invention can be implemented in various forms, and therefore the present invention is not limited to the embodiments described below. In addition, in order to more clearly describe the present invention, components that are not related to the present invention will be omitted from the drawings.

[0035] The technical solutions in the embodiments of the present application are to solve the problems of the above-mentioned background technology, which are generally as follows:

[0036] Example:

[0037] This embodiment introduces a flexible control and optimized dispatching system for a power grid with distributed energy access, including a dynamic perception module, a prediction and correction module, a flexible dispatching module, an adaptive control module, and a collaborative traceability module.

[0038] 1. Dynamic Perception Module

[0039] The dynamic perception module is used to capture output fluctuations and anomalies of distributed energy sources (photovoltaic, wind power, and energy storage) in real time and trigger a hierarchical processing mechanism, providing an accurate data foundation for subsequent forecasting, scheduling, and control, ensuring that the system can respond promptly to changes in energy output. This module includes a high-frequency acquisition unit, an anomaly location unit, and a data repair unit.

[0040] 1. High-frequency acquisition unit

[0041] The function of the high-frequency acquisition unit is to start high-frequency acquisition when the output of distributed energy changes abnormally, so as to obtain more intensive real-time data and support the rapid response of the system. When the output change rate of distributed energy exceeds ±15%, Formula 1 is satisfied: Among them, P t is the current power, P t-1 The unit starts collecting power data in seconds for the previous time series. If data exceeds the limit, it sends a warning signal to the prediction and correction module, indicating that the prediction model may need to be adjusted to accommodate rapid changes in energy output. This way, the high-frequency data collection unit can promptly capture dramatic fluctuations in energy output, providing more accurate data for subsequent system processing.

[0042] The decision to switch the acquisition frequency is calculated in real time using Formula 1. If the rate of change is less than 5% for three consecutive sampling periods, it is determined to be in a stable state, and the system automatically restores the default minute-level acquisition frequency. If the standard deviation of the collected data exceeds twice the standard deviation of the historical data, the system determines that the data is abnormal and triggers the abnormal location unit for in-depth investigation. If the data abnormality is confirmed by the abnormal location unit to be caused by a sensor failure, the system will automatically switch to the backup sensor and send the fault information to the collaborative traceability module for recording. The collaborative traceability module generates a fault log and associates it with the event storage unit.

[0043] 2. Abnormal positioning unit

[0044] The abnormality location unit distinguishes between instantaneous disturbances and equipment failures based on the three time series data of t-1, t, and t+1, so that the system can take corresponding treatment measures for different abnormal situations. When formula 2 is satisfied: If P t+1 ≥0.9P t And P t ≤0.5P r , where P r If the power output is greater than the rated power of the device, it is considered a device failure. Once the device failure is flagged, the adaptive control module triggers the backup power supply switch, ensuring system power reliability. This precise fault determination mechanism enables the anomaly location unit to accurately identify device failures, avoiding misjudgments caused by transient disturbances and ensuring stable system operation.

[0045] The device fault judgment adopts a dual threshold verification mechanism. When the conditions of Formula 2 are met, the system further verifies the auxiliary parameters such as the device temperature and current. If the auxiliary parameters exceed the normal range at the same time, specifically, the temperature exceeds 75°C or the current exceeds 120% of the rated value, it is confirmed as a device failure, and the backup power supply switching of the adaptive control module is immediately triggered, and the fault information is pushed to the event storage unit of the collaborative traceability module. If the auxiliary parameters are normal, the system determines it to be a transient disturbance, and the data will be marked as suspicious data and sent to the data repair unit for processing. For suspicious data, the system will continue to monitor in the subsequent several sampling cycles. If the abnormal situation disappears, the data will be restored to normal data; if the abnormal situation persists, the fault judgment process will be re-triggered.

[0046] 3.Data repair unit

[0047] The data repair unit uses a sliding average filter method to smooth the data for instantaneous disturbances to improve the quality and availability of the data. Specifically, it is calculated through formula 3: in, is the repaired data, P t-1 、P t 、P t+1 The data is repaired for the power at times t-1, t, and t+1. The repaired data is then fed back into the flexible scheduling module in real time, providing more accurate data support for scheduling plans. The data repair unit effectively removes the impact of transient disturbances on the data, making it smoother and more stable, thereby improving the system's scheduling accuracy and reliability.

[0048] The effectiveness of data repair is evaluated by comparing the residual sum of squares (RSS) of the data before and after repair. If the RSS exceeds 1.5 times the historical average RSS, the system determines that the repair has failed, and the data will be marked as invalid and returned to the anomaly location unit for reprocessing. If the deviation between the repaired data and the adjacent time data still exceeds 10%, the system will automatically adjust the sliding window size and use the 5-point sliding average method. The formula is: Perform secondary repair. If the data still does not meet the requirements after the secondary repair, it will be considered unrepairable data. The system will activate the backup data source, specifically the data collected by the adjacent nodes for data supplementation, and record the data anomaly event in the event storage unit of the collaborative traceability module.

[0049] 2. Prediction Correction Module

[0050] The prediction correction module dynamically corrects the prediction model through time series deviation propagation analysis, ensuring that scheduling instructions are consistent with actual conditions. This allows the system to adjust prediction results based on actual energy output, thereby formulating more reasonable scheduling plans. This module includes a deviation analysis unit, a model update unit, and an instruction backtracking unit.

[0051] 1. Deviation Analysis Unit

[0052] The deviation analysis unit is used to calculate the forecast deviation and determine whether it is a trend deviation, providing a basis for updating the model and adjusting the scheduling plan. t =|P pr -P ac |, where P pr is the predicted power, P ac is the actual power. If formula 4 is satisfied for two consecutive times: e t >0.1P r and e t+1 >e t , where P r If the power output is less than the rated power of the equipment, a trend deviation is identified, requiring a global model update. This trend deviation triggers the flexible scheduling module to reset the scheduling plan to adapt to the long-term trend of energy output. Accurate judgment by the deviation analysis unit can promptly identify deviations between the predicted model and actual conditions, providing important evidence for model updates and scheduling plan adjustments.

[0053] Trend deviation judgment adopts a dual threshold trigger mechanism. When the prediction deviation exceeds 0.15P for three consecutive times, r Or the cumulative deviation exceeds 0.5P rWhen the system determines a severe trend deviation, it immediately triggers the global model update process and notifies the flexible scheduling module to reset the scheduling plan. If only the conditions in Formula 4 are met, the system determines a mild trend deviation and adjusts only the local model parameters. For mild trend deviations, the system continuously monitors the deviation changes in subsequent forecast cycles. If the deviation gradually decreases, the current adjustment is maintained; if the deviation continues to increase, it is upgraded to a severe trend deviation to ensure that the forecast model is consistent with the actual output trend.

[0054] 2. Model update unit

[0055] The model update unit dynamically adjusts the meteorological weight of the LSTM model based on the single-point deviation to improve the accuracy and adaptability of the prediction model. Specifically, through formula 5: w n =w o -αδ, where w n is the new weight, w o The weights are adjusted using the original weights, α is the learning rate, and δ is the bias value. The LSTM inputs are historical power, temperature, and irradiance, and the output is the predicted value for the next 15 minutes. The updated model parameters are synchronized to all edge nodes, ensuring a consistent and accurate prediction model across the entire system. Dynamic adjustments to the model update unit enable the prediction model to better adapt to varying meteorological conditions and energy output.

[0056] The effectiveness of model updates is evaluated using the root mean square error (RMSE) of the cross-validation set. If the RMSE after an update decreases by less than 5% compared to before the update, the system deems the update ineffective and rolls back to the previous model parameters. It then initiates a hyperparameter optimization process, re-adjusting the learning rate α and the number of hidden layer neurons using a grid search method. If three consecutive model updates fail to achieve the expected results, the system triggers the root cause analysis unit of the collaborative traceability module, linking meteorological data with equipment operating status for in-depth analysis. The results are then pushed to the model update unit to optimize prediction weights, forming a closed-loop optimization mechanism.

[0057] 3. Instruction backtrace unit

[0058] When a deviation causes a scheduling instruction to fail, the instruction backtracking unit backtracks to the starting point of the deviation and regenerates optimized instructions to compensate for any system performance issues caused by the deviation. If a deviation causes a scheduling instruction to fail, the unit backtracks to the starting point of the deviation, t0, and regenerates optimized instructions from t0 to the current point. The backtracking results are pushed to the adaptive control module to compensate for the output shortfall and ensure stable system operation. The instruction backtracking unit's processing can promptly correct scheduling instruction failures caused by prediction deviations, improving system stability.

[0059] The command failure judgment is based on the power balance error E p =|P sc -Pac |, where P sc is the dispatching power, P ac is the actual power, when E p If the power exceeds 20% of the planned power for the current period and lasts for more than 5 minutes, the system determines that the instruction is invalid. The backtracking depth is determined by Formula 8: D = min(t-t0, 30 minutes) to ensure that the root cause of the problem can be traced within a reasonable time frame. If the new instruction generated after backtracking still cannot meet the power balance requirements, that is, the error still exceeds 15%, the system will expand the backtracking range and perform secondary optimization in conjunction with the flexible scheduling module. The optimization results are synchronized to the adaptive control module for execution, and the backtracking process is recorded in the event storage unit of the collaborative traceability module.

[0060] 3. Flexible Scheduling Module

[0061] The flexible dispatch module generates dispatch instructions based on multi-objective optimization and dynamically responds to fluctuations in distributed energy resources to achieve economical operation and stable power supply. This module includes a cost optimization unit, a voltage stabilization unit, and a real-time correction unit.

[0062] 1. Cost Optimization Unit

[0063] The cost optimization unit aims to minimize the power purchase cost and network loss, and supports the economic operation of the system. Specifically, it uses Formula 6: Among them C g is the grid electricity price, P g The amount of electricity purchased from the power grid, I t is the branch current, R i Optimization is performed, assuming β is the resistance and β is the loss weight. The optimization result serves as the power baseline for the adaptive control module, guiding its operation and achieving economical system operation. Through the optimization calculations of the cost optimization unit, it is possible to minimize power purchase costs and network losses while meeting the system's power supply requirements, thereby improving the system's economic benefits.

[0064] Optimization results are evaluated using a feasibility verification matrix. If the optimized solution causes voltage violations at more than three nodes (as determined by Equation 7) or if the reserve capacity falls below 15%, the system deems the solution infeasible, automatically adds voltage stability constraints, and reoptimizes. If an infeasible solution is identified, the system prioritizes voltage stability, appropriately increasing the power purchase cost to meet the voltage constraints. The reasons for the infeasible solution are recorded in the collaborative traceability module's event storage unit, providing data support for subsequent optimization algorithms.

[0065] 2. Voltage stabilization unit

[0066] The voltage stabilization unit constrains the node voltage deviation to be within a reasonable range to ensure the voltage stability of the system. The specific constraint condition is Formula 7: |V i-V0|≤0.05V0, where V i is the node voltage, and V0 is the reference voltage. When the voltage exceeds the limit, the dynamic perception module triggers a node monitoring upgrade, strengthening monitoring of the node to promptly detect and address voltage issues. Through the constraints and control of the voltage stabilization unit, the system voltage is kept within the normal range, improving the system's power supply quality and stability.

[0067] The voltage exceeds the limit and adopts a three-level warning mechanism: (1) Warning threshold: 0.97V0≤V i ≤1.03V0, triggering the high-frequency acquisition unit of the dynamic perception module to increase the monitoring frequency of the node to the second level; (2) Alarm threshold: 0.95V0≤V i ≤1.05V0, start the reactive power compensation device of the adaptive control module; (3) Emergency threshold: V i <0.95V0 or V i >1.05V0, the adaptive control module is triggered to implement load shedding and record the voltage over-limit event in the event storage unit of the collaborative traceability module. If the voltage over-limit event lasts for more than 10 minutes, the system automatically triggers the root cause analysis unit of the collaborative traceability module, correlating the distributed power output data with the network topology to analyze the cause of the over-limit event.

[0068] 3. Real-time correction unit

[0069] The real-time correction unit updates the dispatch plan every five minutes to respond to real-time output changes, allowing the dispatch plan to promptly adapt to fluctuations in energy output. Correction instructions are sent to edge execution terminals in real time to ensure timely execution of dispatch instructions. This rolling update of the real-time correction unit ensures that the dispatch plan is more closely aligned with actual energy output conditions, improving the system's dispatch accuracy and responsiveness.

[0070] The necessity of correction is determined by the cumulative amount of forecast deviation Evaluation, if E c If the power exceeds 8% of the planned power for the current period, the correction process is triggered. The correction amplitude is calculated by formula 10: ΔP = min (0.2P sc ,E c ) limits to prevent over-correction from causing system oscillation. If the revised scheduling plan performs poorly—specifically, if the power balance error still exceeds 10%—the system shortens the correction cycle to 2 minutes, increases the frequency of corrections, and simultaneously triggers the deviation analysis unit in the prediction correction module to reassess the accuracy of the prediction model, forming a collaborative optimization mechanism for scheduling and prediction.

[0071] 4. Adaptive Control Module

[0072] The adaptive control module maintains grid frequency and voltage stability through parameter adaptation and multi-level response strategies, ensuring safe and stable system operation. This module includes a VSG frequency regulation unit, an island switching unit, and a damping adjustment unit.

[0073] 1. Real VSG FM unit

[0074] The VSG frequency modulation unit simulates the inertia response of synchronous machines to maintain grid frequency stability. This inertia response simulation is achieved through a control algorithm. Frequency anomaly data is fed back to the flexible dispatch module to adjust constraints, enabling the module to adjust the dispatch plan based on frequency fluctuations to maintain system frequency stability. The VSG frequency modulation unit's simulated inertia response improves the system's ability to suppress frequency fluctuations, ensuring that the system's frequency remains within a stable range.

[0075] The frequency regulation effect is evaluated using the rate of change of frequency (ROCOF). When |df / dt| exceeds 0.5 Hz / s, the system identifies a frequency emergency and automatically increases the VSG inertia parameter J to 1.5 times the rated value. This frequency anomaly data is fed back to the flexible dispatch module, triggering it to adjust the active power dispatch constraints. If the frequency deviation Δf exceeds ±0.2 Hz for 30 seconds, the system will coordinate control with the damping control unit to enhance system stability. The coordinated control strategy is recorded in the event storage unit of the collaborative traceability module. Once the frequency returns to the normal range, the system will gradually restore the inertia parameter to its normal value to avoid unnecessary impact on the system.

[0076] 2. Island switching unit

[0077] The island switching unit implements island operation switching in the event of a grid fault, ensuring reliable power supply to critical loads. If the voltage drops by more than 30%, the system switches to island operation within 200ms and supplies power according to the black start plan. This islanding event triggers the collaborative traceability module to record all operations, facilitating subsequent analysis and resolution of the islanding event. The rapid switching of the island switching unit ensures continuous power supply to critical loads during grid faults, improving system power supply stability.

[0078] The island switching uses a multi-criteria fusion algorithm and is triggered when the following conditions are met simultaneously: (1) voltage drop > 30%; (2) frequency deviation |Δf| > 0.5Hz; (3) fault detection time < 100ms. After the switching is successful, the system monitors the frequency and voltage in the island in real time through the synchronized phasor measurement unit (PMU). If the fluctuation exceeds ±0.3Hz or ±5%, the load shedding strategy is activated and the island operation data is transmitted in real time to the event storage unit of the collaborative traceability module. During the island operation, the system will continuously evaluate the stability and power supply capacity of the island. If it is found that stable operation cannot be maintained, it will trigger the synchronization process with the main grid, and the grid connection request will be sent to the flexible scheduling module for grid connection strategy optimization.

[0079] 3. Damping adjustment unit

[0080] The damping adjustment unit dynamically adjusts the D value based on the Lyapunov criterion to improve the stability of the system. Specifically, through formula 9: D n =D0+k·dΔf, where D n The system performs an adjustment based on the new damping coefficient, D0, the original damping coefficient, k, and Δf, the frequency deviation. The adjustment results are synchronized with the prediction and correction module to update the stability threshold. This allows the prediction and correction module to adjust the prediction model parameters based on changes in system stability to improve prediction accuracy. Dynamic adjustments to the damping adjustment unit improve the system's damping characteristics, suppress system oscillations, and ensure stable operation.

[0081] The damping effect is evaluated through system eigenvalue analysis. If the real part of the dominant eigenvalue is greater than -0.8, the system is deemed underdamped and automatically increases the D value until the real part falls within the range of -1.2 to -0.8. A gradient descent method is used to find the optimal D value during the adjustment process, with the step size η adaptively adjusted using Formula 11: η = 0.1·|Re(λ)+1|. If the system still oscillates after damping adjustment, the system will coordinate control with the VSG frequency modulation unit to improve system stability. The coordinated control process is recorded in the event storage unit of the collaborative traceability module, triggering the prediction and correction module to reassess the stability threshold.

[0082] 5. Collaborative traceability module

[0083] The collaborative traceability module records the entire system operation chain and supports fault backtracking and policy optimization, allowing the system to quickly locate the cause of a fault and optimize the system's policies and parameters. This module includes an event storage unit, a root cause analysis unit, and a policy self-update unit.

[0084] 1. Event storage unit

[0085] The event storage unit stores data from the entire "perception-dispatching-control" chain by timestamp, supporting millisecond-level retrieval and providing data support for fault tracing and system analysis. Data anomalies trigger the root cause analysis unit to enable timely analysis of the cause of the anomaly. The detailed data records in the event storage unit provide accurate data support for system fault analysis and optimization.

[0086] Data integrity is verified through a hash check mechanism. If the hash value of a timestamp data is inconsistent with the pre-calculated value, the system determines that the data is corrupted and immediately starts the data recovery process, giving priority to obtaining it from the edge node cache, and triggering the root cause analysis unit to locate the source of data corruption. At the same time, the system will evaluate the continuous integrity of the stored data. If more than 5 data points are lost continuously, an early warning will be triggered and the root cause analysis unit will be notified to associate the network communication status for fault location. For damaged data, the system will attempt to restore it from the backup data source; if it cannot be restored, it will be marked as invalid data and the impact of the data will be excluded from subsequent analysis. If the data corruption is caused by a hardware failure, the root cause analysis unit will push the fault information to the adaptive control module, trigger the switching of the backup storage path, and generate a hardware fault log in the event storage unit, which will be associated with the equipment status monitoring data to provide a basis for subsequent maintenance.

[0087] 2. Root Cause Analysis Unit

[0088] The root cause analysis unit locates the source of faults when data anomalies occur, providing a basis for troubleshooting and system optimization. If the same node triggers the same anomaly three times within a week, the source of the fault is identified through correlation with meteorological and equipment status data. The analysis results are then sent to the model update unit to optimize prediction weights, allowing the prediction model to better adapt to system operating conditions. Accurate analysis by the root cause analysis unit allows for rapid locating of fault sources, improving system troubleshooting efficiency.

[0089] Root cause location uses Bayesian network reasoning to calculate conditional probability Where F is the fault type and E is the abnormal event, the most likely fault source is determined. When the posterior probability of a fault type, P(F|E), is greater than 0.75, the system identifies it as the root cause and generates a maintenance work order that is pushed to the execution terminal of the adaptive control module. At the same time, the root cause analysis results are synchronized to the policy self-update unit as a basis for policy optimization. If a unique root cause cannot be determined, that is, the posterior probabilities of multiple fault types are greater than 0.5, the system will expand the analysis scope and call the high-frequency acquisition unit of the dynamic perception module to perform encrypted sampling on the relevant nodes. After supplementing the data, Bayesian inference will be re-performed. If the root cause location fails twice in a row, the system will trigger the policy self-update unit and automatically switch to the alternative analysis strategy for multi-source faults.

[0090] 3. Policy self-update unit

[0091] The policy self-update unit automatically replaces a policy with an alternative solution when it fails, improving system adaptability and reliability. If a policy fails within five invocations, it is automatically replaced with an alternative solution and distributed to edge devices. After verification, the new policy is synchronized to all modules, ensuring optimal policy across the entire system. Automatic updates from the policy self-update unit ensure that the system consistently adopts the optimal strategy in varying operating conditions, improving system efficiency and adaptability.

[0092] Strategy failure is assessed by the magnitude of a decrease in key performance indicators (KPIs). If a strategy causes a frequency deviation exceeding ±0.2Hz or a voltage limit increase of more than 20% for three consecutive times, the system deems the strategy failed. Alternative strategies are selected using a multi-objective decision matrix, with comprehensive evaluation criteria including response speed (40%), stability improvement (35%), and cost increase (25%). New strategies are piloted and validated on edge nodes for 48 hours. If the KPI improves by more than 15% during this period, they are rolled out across the network. If the improvement is between 5% and 15%, the system adjusts the strategy parameters and extends the validation period to 72 hours. If the improvement is less than 5%, the new strategy is deemed invalid, reverting to the previously valid strategy, and reselecting from the alternative strategy library. After a new strategy is implemented, the event storage unit continuously tracks its effectiveness. If it fails three times in 10 subsequent invocations, the system automatically marks the strategy as "restricted use" and initiates a new round of strategy optimization.

[0093] Finally, it should be noted that the above embodiments are merely examples for the purpose of illustrating the present invention and are not intended to limit the embodiments. Those skilled in the art will readily appreciate that other variations or modifications based on the above description are possible. It is not necessary and impossible to provide an exhaustive list of all embodiments. However, obvious variations or modifications arising therefrom remain within the scope of protection of the present invention.

Claims

1. A flexible control and optimized dispatching system for distributed energy access, characterized by: It includes dynamic perception module, prediction and correction module, flexible scheduling module, adaptive control module and collaborative traceability module; The dynamic perception module includes a high-frequency acquisition unit, an anomaly location unit, and a data repair unit, which are used to capture the output fluctuations and anomalies of distributed energy in real time and trigger a hierarchical processing mechanism to provide data for subsequent prediction, scheduling, and control; The prediction correction module includes a deviation analysis unit, a model update unit, and an instruction backtracking unit, which is used to dynamically correct the prediction model through timing deviation propagation analysis to ensure that the scheduling instructions are in line with the actual situation; The flexible scheduling module includes a cost optimization unit, a voltage stabilization unit, and a real-time correction unit, which are used to generate scheduling instructions based on multi-objective optimization and dynamically respond to fluctuations in distributed energy resources; The adaptive control module includes a VSG frequency regulation unit, an island switching unit, and a damping adjustment unit, which are used to maintain grid frequency and voltage stability through parameter adaptation and multi-level response strategies; The collaborative tracing module includes an event storage unit, a root cause analysis unit, and a policy self-update unit, which are used to record the entire system operation chain and support fault backtracking and policy optimization.

2. The grid flexible control and optimization scheduling system for distributed energy access according to claim 1, characterized in that: The high-frequency acquisition unit of the dynamic sensing module starts second-level acquisition when the output change rate of distributed energy exceeds 15%, and if the change rate is less than 5% within three consecutive sampling periods, the default minute-level acquisition frequency is restored; The abnormality location unit distinguishes between transient disturbance and equipment failure based on three time series data. t+1 ≥0.9P t And P t ≤0.5P r When, P r is the rated power of the equipment, P t If it is the current power, it is determined to be a device failure; otherwise, it is determined to be a transient disturbance; The data repair unit uses a sliding average filtering method to repair instantaneous disturbances, and evaluates the repair effect by comparing the residual sum of squares of the data before and after repair. If the repair fails, the sliding window size is adjusted or a backup data source is started.

3. The grid flexible control and optimization scheduling system for distributed energy access according to claim 1, characterized in that: The deviation analysis unit of the prediction correction module is used to calculate the prediction deviation and determine whether it is a trend deviation. t >0.1P r and e t+1 >e t When, the prediction error e t =|P pr -P ac |,P pr is the predicted power, P ac is the actual power, P r The rated power of the equipment is determined to be a trend deviation and requires a global model update; The model updating unit dynamically adjusts the meteorological weight of the LSTM model for single-point deviations and evaluates the update effect through the root mean square error of the cross-validation set. If the update effect is not good, the model parameters are rolled back and the hyperparameter optimization process is started; When the deviation causes the scheduling instruction to fail, the instruction backtracking unit backtracks to the deviation starting point and regenerates the optimization instruction, and pushes the backtracking result to the adaptive control module.

4. The grid flexible control and optimization scheduling system for distributed energy access according to claim 1, characterized in that: The cost optimization unit of the flexible scheduling module optimizes with the goal of minimizing the power purchase cost and network loss, and evaluates the optimization results through a feasibility verification matrix. If the solution is not feasible, it automatically adds constraints and re-optimizes; The voltage stabilization unit constrains node voltage deviations within a reasonable range and uses a three-level early warning mechanism to trigger corresponding operations of the dynamic perception module and the adaptive control module; The real-time correction unit updates the scheduling plan every 5 minutes and evaluates the necessity of correction based on the accumulated prediction deviation. If the execution effect after correction is not good, the correction cycle is shortened and the prediction correction module is triggered to re-evaluate the accuracy of the prediction model.

5. The grid flexible control and optimization scheduling system for distributed energy access according to claim 1, characterized in that: The VSG frequency regulation unit of the adaptive control module simulates the inertia response of the synchronous machine through the control algorithm to maintain the stability of the grid frequency, and evaluates the regulation effect through the frequency change rate. When the frequency is abnormal, it feeds back to the flexible scheduling module to adjust the constraint conditions; The island switching unit triggers island operation switching according to a multi-criteria fusion algorithm when a power grid fault occurs, and monitors the frequency and voltage in the island in real time through the PMU; The damping adjustment unit dynamically adjusts the damping coefficient based on the Lyapunov criterion and evaluates the damping effect through system eigenvalue analysis. If oscillation still exists after the damping adjustment, the VSG frequency modulation unit is combined for coordinated control.

6. The grid flexible control and optimization scheduling system for distributed energy access according to claim 1, characterized in that: The event storage unit of the collaborative traceability module stores the "perception-scheduling-control" full-link data by timestamp and verifies the data integrity through a hash check mechanism; The root cause analysis unit uses Bayesian network reasoning to locate the fault source when data is abnormal, and pushes the analysis results to the model update unit and the strategy self-update unit; The strategy self-update unit automatically replaces the strategy with an alternative plan when the strategy fails, and evaluates the strategy failure through key performance indicators. The new strategy is synchronized to all modules after verification.

Images