A new energy power station power generation curve management system based on power intelligent scheduling

Abstract

This invention discloses a power generation curve management system for new energy power plants based on intelligent power dispatch, belonging to the field of new energy management technology. It includes an information acquisition module for acquiring dynamic adjustment capability information of the power grid in a future dispatch cycle, and for acquiring the predicted power generation curve of the new energy power plant in the future dispatch cycle. The dynamic adjustment capability information includes time-varying adjustment rate constraints and adjustment capacity constraints. A fluctuation decomposition module is used to decompose the predicted power generation curve into followable fluctuation components and non-followable fluctuation components according to the adjustment rate constraints. This invention, through the identification of capacity relaxation windows and spatiotemporal translation operations, realizes a fundamental transformation of the new energy power generation curve from passively responding to real-time grid constraints to actively adapting to the grid's adjustment rhythm, resolving the core contradiction of mismatch between the power generation curve and the grid's acceptance capacity rhythm.

Description

Technical Field

[0001] This invention belongs to the field of new energy management technology, specifically a new energy power plant power generation curve management system based on intelligent power dispatching. Background Technology

[0002] The intermittent and fluctuating characteristics of renewable energy power plants have become a core bottleneck restricting their large-scale grid connection and absorption. Current renewable energy power generation curve management often employs fixed-threshold peak-shaving and valley-filling strategies, performing local optimization based solely on the grid's regulation capacity at a single moment, without fully considering the time-varying characteristics of the grid's dynamic regulation capabilities. Specifically, existing technologies generally ignore the impact of regulation rate constraints on the power generation curve, treating the predicted power generation curve as a whole. This results in some fluctuating components exceeding the grid's regulation rate and failing to be effectively absorbed, leading to power curtailment or grid frequency fluctuations.

[0003] Meanwhile, existing solutions lack a precise identification mechanism for periods of grid capacity relaxation, fail to redistribute unadjustable fluctuation components in time and space, and do not consider the coordinated participation of energy storage systems and demand-side response resources. They can only deal with the power exceeding the constraints through hard reduction, resulting in a waste of the potential of new energy power generation.

[0004] In addition, traditional systems do not establish a matching relationship between the fluctuation pattern and the grid's dynamic adjustment capability when generating the target power generation curve. This results in the adjusted power generation curve still being incompatible with the grid's acceptance capacity, affecting the grid's operational stability and the efficiency of new energy consumption. Summary of the Invention

[0005] The purpose of this invention is to provide a power generation curve management system for new energy power plants based on intelligent power dispatching, so as to solve the problems mentioned in the background art.

[0006] A power generation curve management system for new energy power plants based on intelligent power dispatching includes: The information acquisition module is used to acquire dynamic regulation capability information of the power grid in the future dispatch cycle, and to acquire the predicted power generation curve of the new energy power station in the future dispatch cycle; the dynamic regulation capability information includes regulation rate constraints and regulation capacity constraints that change over time. The fluctuation decomposition module is used to decompose the predicted power generation curve into a followable fluctuation component and a non-followable fluctuation component according to the adjustment rate constraint; wherein, the followable fluctuation component refers to the fluctuation part whose rate of change never exceeds the adjustment rate constraint at the corresponding time, and the non-followable fluctuation component refers to the fluctuation part whose rate of change exceeds the adjustment rate constraint at the corresponding time. The window identification module is used to determine at least one capacity relaxation window within the future scheduling cycle based on the time-varying characteristics of the adjustment capacity constraint; the capacity relaxation window refers to a continuous period of time when the adjustment capacity constraint value is high and the power grid has additional absorption capacity. The curve generation and translation module is used to separate the unfollowable fluctuation component from the original time period and translate it as a whole into the capacity relaxation window for centralized release, so as to generate a target power generation curve. The fluctuation pattern of the target power generation curve matches the spatiotemporal distribution of the grid's dynamic adjustment capability. The control execution module is used to control the power generation unit of the new energy power station according to the target power generation curve.

[0007] In some possible implementations, the window recognition module is used for: Based on the prediction of future regulation capacity change trends, the curve of the regulation capacity constraint changing over time is analyzed to identify the complete fluctuation cycle of regulation capacity from low to high and then back down. In each fluctuation cycle, the continuous period when the regulation capacity is at a high level is identified as the candidate window; Based on the time span and energy magnitude of the unfollowable fluctuation component, a window that can fully accommodate the unfollowable fluctuation component is selected from the candidate windows and used as the capacity relaxation window.

[0008] In some possible implementations, the curve generation and translation module is used for: Based on the start time and duration of the capacity relaxation window, the unfollowable fluctuation component is shifted as a whole on the time axis so that it falls completely within the capacity relaxation window, and the peak time of the unfollowable fluctuation component is aligned with the peak time of the adjusted capacity within the capacity relaxation window. Within the capacity relaxation window, the new energy power plant is controlled to generate electricity at a rate exceeding the regulation rate constraint, so as to utilize the peak absorption capacity of the power grid during this period to absorb the unfollowable fluctuation component. Outside the capacity relaxation window, the output curve of the new energy power plant is controlled to remain consistent with the followable fluctuation component, so that its output speed is always limited by the adjustment rate constraint.

[0009] In some possible implementations, the fluctuation decomposition module is used for: The predicted power generation curve is compared point by point with the regulation rate constraint; The fluctuation segments whose rate of change exceeds the adjustment rate constraint are marked, and the fluctuation segments are spliced ​​together to form the unfollowable fluctuation component; The followable fluctuation component is obtained by subtracting the non-followable fluctuation component from the predicted power generation curve.

[0010] In some possible implementations, the fluctuation decomposition module is further configured to: The continuous fluctuations in the predicted power generation curve are divided into several independent fluctuation events, each with a clear start point, peak point, and end point; Each fluctuation event is determined to be fully accommodated by the adjustment rate constraint. Fluctuations that cannot be accommodated are marked as the unfollowable fluctuation components.

[0011] In some possible implementations, a coordinated absorption module is also included, used when there is no capacity relaxation window within the future scheduling period that can fully accommodate the unfollowable fluctuation component: Calculate the energy gap between the total energy of the non-followable fluctuation component and the remaining absorption capacity of the capacity relaxation window; The energy gap is allocated to energy storage systems and demand-side response resources to generate coordinated consumption instructions; According to the coordinated absorption command, the energy storage system is activated for charging and discharging, and the demand-side load is adjusted to assist in the absorption of the remaining portion of the non-followable fluctuation component.

[0012] In some possible implementations, a verification module is also included, used after the curve generation and translation module has translated the unfollowable fluctuation component to the capacity relaxation window: The target power generation curve is dynamically compatibility verified; The dynamic compatibility verification includes: verifying whether the maximum output speed of the shifted non-followable fluctuation component within the capacity relaxation window exceeds the allowable upper limit of the regulation rate constraint for that period, and verifying whether the cumulative power generation of the target power generation curve at any time exceeds the regulation capacity constraint for the corresponding time. If the verification passes, the target power generation curve is output; if the verification fails, the translation position is adjusted or the backup absorption mechanism is activated.

[0013] In some possible implementations, the dynamic adjustment capability information includes the adjustment response speed, adjustment duration, and adjustment cost of different types of power supplies; The regulation rate constraint and the regulation capacity constraint are generated through optimization calculations based on the regulation response speed, regulation duration, and regulation cost, so that the generated constraints minimize the overall regulation cost while ensuring power grid security.

[0014] In some possible implementations, the curve generation and translation module is further configured to reshape the waveform of the unfollowable fluctuation component after translating it into the capacity relaxation window: Obtain the adjustment rate constraint change curve within the capacity relaxation window; Based on the adjustment rate constraint change curve, the internal morphology of the shifted non-followable fluctuation component is adjusted so that the instantaneous rate of change of the adjusted non-followable fluctuation component does not exceed the adjustment rate constraint value at any time within the capacity relaxation window, while keeping the total energy of the non-followable fluctuation component constant.

[0015] In some possible implementations, the curve generation and translation module, when adjusting the internal shape of the translated non-followable fluctuation component, is used to: Identify the original waveform characteristics of the unfollowable fluctuation component, the original waveform characteristics including the steepness of the rising segment, the peak amplitude, and the steepness of the falling segment; Based on the changing trend of the adjustment rate constraint within the capacity relaxation window, the unfollowable fluctuation component is divided into multiple sub-segments; The multiple sub-segments are reordered and spliced ​​within the capacity relaxation window, so that the sub-segments with a steeper rising segment correspond to the time period with a higher adjustment rate constraint, and the sub-segments with a steeper falling segment correspond to the time period with the second highest adjustment rate constraint, and the re-spliced ​​waveform remains continuous and the total energy remains unchanged.

[0016] The technical solutions provided by the embodiments of this disclosure have at least the following beneficial effects: By identifying the "capacity relaxation window" and performing "spatiotemporal translation" operations, a fundamental shift was achieved in the new energy power generation curve from passively responding to real-time grid constraints to actively adapting to the grid's adjustment rhythm, thus resolving the core contradiction of mismatch between the power generation curve and the grid's acceptance capacity rhythm.

[0017] By shifting unfollowable fluctuation components to periods when the grid has sufficient absorption capacity, energy that was originally forced to be abandoned due to insufficient regulation capacity can be effectively absorbed. Without increasing rigid investment, the implicit absorption potential of the grid is released, and the utilization rate of new energy sources is improved.

[0018] By using waveform reshaping and peak alignment mechanisms, it is ensured that the rate of change of the shifted fluctuation component at any time within the window does not exceed the regulation rate constraint. At the same time, the peak fluctuation is precisely matched with the peak regulation capacity of the power grid. Under the premise of ensuring the safe and stable operation of the power grid, the peak absorption capacity of the power grid is maximized and the absorption efficiency is improved. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the system framework structure of the present invention. Detailed Implementation

[0020] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see Figure 1 This application provides a power generation curve management system for new energy power plants based on intelligent power dispatching, comprising: An information acquisition device is used to acquire information on the dynamic regulation capability of the power grid in a future dispatch cycle, and to acquire the predicted power generation curve of the new energy power station in a future dispatch cycle; the dynamic regulation capability information includes time-varying regulation rate constraints and regulation capacity constraints.

[0022] The dynamic adjustment capability information includes the adjustment response speed, adjustment duration, and adjustment cost for different types of power supplies.

[0023] The regulation rate constraint and the regulation capacity constraint are generated through optimization calculations based on the regulation response speed, regulation duration, and regulation cost, so that the generated constraints minimize the overall regulation cost while ensuring power grid security.

[0024] Specifically, the duration of the future dispatch cycle can be set according to the grid dispatch needs. In this embodiment, the value is 24 hours. This value is determined based on the daily load cycle change pattern of the power grid, which can fully cover the full-time characteristics of load peak, flat and low periods, and is in line with the industry's conventional mode of daily dispatch of the power system.

[0025] The time granularity is 15 minutes, meaning the entire scheduling cycle is divided into 96 time nodes. This granularity is determined based on the balance between scheduling accuracy and computational load. It meets the real-time response requirements of the power grid's dynamic regulation capability without causing a surge in data volume and waste of computational resources due to excessively fine granularity. At the same time, it is compatible with the conventional sampling frequency of the 15-minute statistical average of the SCADA system for new energy power plants.

[0026] The power grid dynamic regulation capacity information is obtained through the real-time data interaction interface of the power grid dispatch center. Data transmission is achieved using the GB / T19582-2018 standard communication protocol. The transmission frequency and time granularity are kept consistent, that is, the regulation rate constraint and regulation capacity constraint data are updated every 15 minutes.

[0027] Different types of power sources include thermal power, hydropower, energy storage, wind power, photovoltaic power, and demand-side response resources. The information acquisition device collects data on the regulation response speed, regulation duration, and regulation cost of various power sources in real time through the dispatch master station. The regulation response speed represents the time delay from receiving the command to the change in output power of the power source. The regulation duration represents the longest time that the power source can maintain the target regulation power. The regulation cost represents the economic cost generated per unit of regulation power.

[0028] The optimization calculation of the regulation rate constraint and the regulation capacity constraint adopts a multi-objective linear programming model, which is constructed as follows: First, the decision variables are determined as the adjustment rate constraint value and adjustment capacity constraint value for each time period in the future scheduling cycle (96 time periods in this embodiment).

[0029] Secondly, the optimization objective is set as minimizing the overall regulation cost throughout the entire cycle. This cost consists of two parts: the cost corresponding to the regulation rate in each time period and the cost corresponding to the regulation capacity in each time period. The unit cost coefficient for each time period is calculated by weighting the regulation response speed, regulation duration, and regulation cost of different types of power sources (thermal power, hydropower, energy storage, etc.).

[0030] Next, set constraints, which mainly include: Power grid safety constraints: Ensure that frequency deviation is controlled within ±0.1Hz, voltage deviation is controlled within ±5%, and power flow on transmission lines does not exceed the limit value in each time period; Regulation capacity constraint: The regulation rate and regulation capacity in each time period cannot exceed the maximum regulation capacity of the adjustable resources in the system; Regulation consistency constraint: Regulation capacity and regulation rate must satisfy a physical relationship. For example, regulation capacity cannot exceed the product of regulation rate and time period length.

[0031] The above optimization model is solved using the simplex method in linear programming. The optimal regulation rate constraint value and regulation capacity constraint value for each time period can be obtained through iterative calculation, which serve as the constraint output for the dynamic regulation capability of the power grid.

[0032] The regulation rate constraint refers to the maximum change in power generation that the power grid can accommodate per unit time, measured in megawatts per minute (MW / min). Its value ranges from 0.5 MW / min to 5 MW / min, and varies dynamically with the grid load level.

[0033] For example, when the grid load factor is higher than 85%, the regulation rate constraint is set at 0.5 to 1.5 MW per minute, and when the load factor is lower than 60%, it is set at 3 to 5 MW per minute. This range is determined based on typical regulation capacity test data of regional power grids in my country, covering the safe operation threshold of the power grid under different operating conditions such as heavy load and light load, to ensure the rationality and practicality of the regulation rate constraint.

[0034] The regulation capacity constraint refers to the maximum additional renewable energy power that the power grid can accept at a specific time. It is measured in megawatts and ranges from 10 to 100 megawatts. It is adjusted in real time according to the grid's reserve capacity. This range refers to the conventional reserve capacity configuration standards for renewable energy consumption in provincial power grids in my country and can adapt to the grid connection needs of renewable energy power plants of different sizes.

[0035] Understandably, the predicted power generation curve of a new energy power plant is generated by a combined prediction model that integrates numerical weather forecast data, historical power generation data of the power plant, and equipment operation status data.

[0036] Furthermore, the combined prediction model adopts a weighted average fusion architecture, in which the weight coefficient of numerical weather prediction data is 0.4, the weight coefficient of historical power generation data of the power plant is 0.35, and the weight coefficient of equipment operation status data is 0.25. This weight allocation is determined by training on historical data of the past 6 months using the least squares method. That is, by minimizing the mean square error between the predicted value and the actual power generation value, the weight ratio of each data source is iteratively optimized. This weight ratio is chosen because numerical weather prediction data has the strongest correlation with short-term power output prediction, followed by historical data. Equipment status data is mainly used to correct deviations caused by sudden failures, which is in line with the technical logic of new energy power generation prediction.

[0037] During model training, if a certain type of data is missing, the weight of that type of data will be automatically distributed to the other two types of data, keeping the total weight sum to 1.

[0038] The numerical weather prediction data is obtained from a third-party meteorological service platform and includes parameters such as wind speed and light intensity, with a time resolution of 15 minutes.

[0039] Historical power generation data is taken from the power plant’s local database and covers records from the same period over the past 12 months.

[0040] Equipment operating status data is collected through the power station's SCADA system, including operating parameters of photovoltaic inverters and wind turbine converters.

[0041] The output form of the predicted power generation curve is a discrete power sequence, with each time node corresponding to a power value. The prediction error is controlled within ±8%, and this error threshold is determined according to the technical requirements for power prediction of new energy power plants in GB / T34120-2017, which meets the industry standard for prediction accuracy of power grid dispatch.

[0042] To ensure data consistency and validity, the information acquisition device performs a data verification process after receiving the data.

[0043] For data on regulation rate constraints and regulation capacity constraints, if the value exceeds the preset reasonable range at a certain moment, or if the data does not change for three consecutive time points, it is determined to be data abnormal. At this time, the system will automatically send a retransmission request to the power grid dispatch center and use historical data from the same period of the previous cycle as a temporary replacement.

[0044] For the predicted power generation curve, if the power value at a certain time point exceeds the range of 0 to 110% of the installed capacity of the power station, the linear interpolation method is used to correct the outlier. The range limit is determined based on the rated output characteristics of the power station equipment. The upper limit of 110% takes into account the short-term overload capacity of new energy equipment, while avoiding the distortion of subsequent decomposition and scheduling logic caused by abnormal data.

[0045] A fluctuation decomposition device is used to decompose the predicted power generation curve into a followable fluctuation component and a non-followable fluctuation component according to the adjustment rate constraint; wherein, the followable fluctuation component refers to the fluctuation part whose rate of change never exceeds the adjustment rate constraint at the corresponding time, and the non-followable fluctuation component refers to the fluctuation part whose rate of change exceeds the adjustment rate constraint.

[0046] Before decomposition, the fluctuation decomposition device is also used to split the continuous fluctuations in the predicted power generation curve into several independent fluctuation events, each fluctuation event having a clear start point, peak point and end point.

[0047] Specifically, the following judgment rules are used to break down fluctuation events: Starting point: When the rate of increase of power for two consecutive time nodes exceeds 20% of the adjustment rate constraint at the corresponding time point, and the power value exceeds 5% of the average power of the previous hour, the first rising node is marked as the starting point.

[0048] Peak point: Starting from the starting point, when the power changes from increasing to decreasing (i.e., the rate of change of two consecutive nodes is negative), or the rate of increase is continuously lower than 10% of the constraint, the last node before turning will be marked as the peak point.

[0049] Termination point: Starting from the peak point, when the power falls back to within ±5% of the starting point power, and the absolute value of the change rate of three consecutive nodes is less than 15% of the constraint, the first node that meets the condition is marked as the termination point.

[0050] Event integrity: The continuous time period between the start point and the end point is an independent fluctuation event, making each fluctuation event independent in time and with clear boundaries, which facilitates subsequent event-by-event judgment on whether it can be accommodated by the power grid regulation capacity.

[0051] The steady state is defined as the absolute value of the power change rate at three consecutive time points being less than 30% of the regulation rate constraint. This ratio is determined based on the need to distinguish between fluctuation events and normal power fluctuations, which can effectively eliminate the interference of small fluctuations and accurately identify fluctuation events with actual absorption impact.

[0052] Each fluctuation event is determined to be fully accommodated by the adjustment rate constraint. Fluctuations that cannot be accommodated are marked as the unfollowable fluctuation components.

[0053] Specifically, for each independent fluctuation event, the power change rate is calculated point by point from the start point to the peak point and from the peak point to the end point, and compared with the regulation rate constraint of the corresponding time period. If the change rate at any point in the fluctuation event exceeds the regulation rate constraint, the fluctuation event is determined to be unacceptable. The entire power range of the fluctuation event from the start point to the end point is marked as an unfollowable fluctuation component. The overall marking method avoids fragmenting individual fluctuation events and ensures the physical integrity and temporal continuity of the unfollowable fluctuation component. This judgment logic is determined based on the veto principle, that is, as long as there is a risk of the fluctuation event exceeding the constraint, it is classified as an unfollowable component to ensure the safe operation of the power grid.

[0054] The fluctuation decomposition device is used to compare the predicted power generation curve with the regulation rate constraint point by point.

[0055] The fluctuation segments whose rate of change exceeds the adjustment rate constraint are marked, and the fluctuation segments are spliced ​​together to form the unfollowable fluctuation component.

[0056] The followable fluctuation component is obtained by subtracting the non-followable fluctuation component from the predicted power generation curve.

[0057] Specifically, point-by-point comparison takes each time node within the scheduling cycle as a unit, sequentially obtains the difference between the predicted power generation of the current node and the previous node, divides it by the node time interval to obtain the power change rate, and then compares the change rate with the adjustment rate constraint at the corresponding time to determine whether it exceeds the constraint range point by point. The point-by-point comparison method can ensure that the entire curve is traversed without omission or deviation, avoiding missed detection of fluctuation segments caused by segmented processing.

[0058] When the power change rate of multiple consecutive nodes exceeds the regulation rate constraint at the corresponding time, the power fluctuation part corresponding to these consecutive nodes is marked as an independent fluctuation segment. All marked fluctuation segments are spliced ​​together in chronological order. The node positions that do not exceed the constraint are filled with zero power, thus forming a non-followable fluctuation component with a complete time series length. The splicing process keeps the time series and energy unchanged, and only retains the part that exceeds the regulation rate constraint, ensuring that the physical meaning of the non-followable fluctuation component is clear and reproducible.

[0059] The followable fluctuation component is obtained by point-to-point numerical subtraction. That is, at the same time node, the power value of the predicted power generation curve is subtracted from the corresponding power value of the non-followable fluctuation component. The power sequence obtained after the subtraction is the followable fluctuation component. The power change rate of this component at all nodes does not exceed the adjustment rate constraint at the corresponding time. This achieves the complete splitting of the original predicted power generation curve into a followable part and a non-followable part, while satisfying the energy conservation relationship.

[0060] In one implementation of this step, the fluctuation decomposition employs a piecewise recursive decomposition algorithm. First, taking the first time node of the future scheduling cycle as the starting point, the initial power value is set as the predicted power generation power of that node.

[0061] Subsequently, each subsequent time node is processed sequentially according to time order, the difference in predicted power between the current node and the previous node is calculated, and then divided by the time interval, which is 15 minutes in this embodiment, or 0.25 hours, to obtain the predicted power change rate.

[0062] Furthermore, the calculated predicted power change rate is compared with the current adjustment rate constraint.

[0063] If the absolute value of the predicted power change rate is less than or equal to the adjustment rate constraint, the predicted power value of the current node is directly assigned to the followable fluctuation component.

[0064] If the absolute value of the predicted power change rate is greater than the adjustment rate constraint, then the adjustment rate constraint is used as the upper limit to calculate the maximum allowable change value of the fluctuation component that the previous node can follow. The difference between this allowable change value and the fluctuation component that the previous node can follow is the maximum value of the fluctuation component that the current node can follow. The difference between the predicted power value of the current node and this maximum value is classified as the non-followable fluctuation component.

[0065] It should be noted that the calculation of the non-followable fluctuation component must adhere to the principle of energy conservation. That is, the sum of the cumulative energy of the followable and non-followable fluctuation components over the entire dispatch cycle must equal the cumulative energy of the original predicted power generation curve. For example, assuming the predicted power at a certain time point is 100 MW, the followable fluctuation component at the previous point is 80 MW, the current adjustment rate constraint is 2 MW per minute, and the time interval is 0.25 hours, then the maximum allowable value of the followable fluctuation component at the current point is 80 MW plus 2 MW per minute multiplied by 0.25 multiplied by 60 minutes, resulting in 110 MW. Since the predicted power of 100 MW is less than 110 MW, the followable fluctuation component at the current point is 100 MW, and the non-followable fluctuation component is 0 MW. If the predicted power at the current point is 120 MW, then the followable fluctuation component is 110 MW, and the non-followable fluctuation component is 10 MW.

[0066] To improve decomposition accuracy, a sliding window smoothing process is introduced during the decomposition. The length of the sliding window is set to three time nodes. This length is determined based on the balance between the smoothing requirements of the fluctuation decomposition and the response speed. If it is too short, it cannot effectively correct single-point fluctuation deviations, and if it is too long, the decomposition results will lag behind the actual power changes. The length of three nodes can achieve the optimal balance between smoothing effect and response speed.

[0067] That is, the decomposition results of each node are corrected by combining the power change trends of the previous and next nodes to avoid decomposition deviations caused by fluctuations in data from a single node.

[0068] Specifically, if the proportion of the non-followable fluctuation component of a certain node exceeds 30% of the predicted power of that node, this proportion threshold is determined based on a large number of simulation tests. When the proportion exceeds 30%, there is a high probability of decomposition deviation or data anomaly. Therefore, it is necessary to recalculate the decomposition results of one time node before and after that node, adjust the distribution ratio of the followable fluctuation component, and ensure the rationality and continuity of the decomposition results.

[0069] A window identification device is used to determine at least one capacity relaxation window within the future scheduling cycle based on the time-varying characteristics of the regulating capacity constraint; the capacity relaxation window refers to a continuous period in which the regulating capacity constraint value is high and the power grid has additional absorption capacity.

[0070] The window identification device is used to analyze the change curve of the regulation capacity constraint over time based on the prediction of the future regulation capacity change trend, identify the complete fluctuation cycle of the regulation capacity from low to high and then back down, determine the continuous period when the regulation capacity is at a high level in each fluctuation cycle as a candidate window, and select the window that can fully accommodate the unfollowable fluctuation component from the candidate window according to the time span and energy magnitude of the unfollowable fluctuation component as the capacity relaxation window.

[0071] It should be understood that the prediction of the trend of capacity change is achieved by using the LSTM time series prediction model. This model is selected based on the mainstream technical solution in the industry for time series data prediction. The LSTM model has the advantages of low gradient vanishing risk and strong long-term dependency capture capability when processing power system time series data, and has higher prediction accuracy than traditional models such as ARIMA.

[0072] The model uses time-series data of the same period of the past 7 days as training samples for the adjustment capacity constraint. The 7-day sample length can cover the weekly load change pattern, while avoiding the low training efficiency caused by excessive sample size.

[0073] The input features include parameters such as time nodes, grid load factor, and reserve capacity ratio. The output is the predicted value of the regulation capacity constraint for each time node in the future scheduling cycle. The prediction error is controlled within ±5%. This error threshold is determined based on the accuracy requirements of the grid scheduling for regulation capacity prediction and can meet the reliability requirements of window identification.

[0074] Specifically, this LSTM model employs a two-layer network structure with 128 and 64 hidden layer units, and an input time window length of 24 time nodes (6 hours). The loss function is mean squared error (MSE), the optimizer is Adam, the initial learning rate is set to 0.001, decaying by a factor of 0.9 every 10 epochs, and the training run consists of 200 epochs. Early stopping is used to prevent overfitting (the model terminates if the validation set loss does not decrease for 20 consecutive epochs).

[0075] The training data are taken from characteristic data such as regulation capacity, grid load factor, and reserve capacity ratio for at least one year of historical data. The data is input into the model after being normalized.

[0076] The test set uses data from the most recent three months, with a sample size of approximately 13,000 time points. The mean absolute percentage error (MAPE) obtained from the test is 4.2%, which meets the prediction accuracy requirement of ±5%.

[0077] The regulation capacity constraint prediction curve output by this model, combined with the actual regulation capacity constraint data obtained by the information acquisition device, forms a complete regulation capacity change curve over time, providing a data basis for identifying fluctuation cycles.

[0078] The identification of a complete fluctuation cycle requires the following change characteristics: from low to high and then back to low, that is, there are clear valley points, peak points and fall points.

[0079] The valley point is defined as the adjustment capacity constraint value at a certain time node being lower than the values ​​of the two adjacent nodes, and lower than 60% of the average adjustment capacity constraint value for the entire scheduling cycle. This proportion is determined based on the statistical characteristics of the adjustment capacity distribution and can effectively distinguish between true low capacity periods and normal fluctuations.

[0080] The peak point is defined as the adjustment capacity constraint value at a certain time node that is higher than the values ​​of the two adjacent nodes and higher than 140% of the average adjustment capacity constraint value of the entire scheduling cycle. This ratio ensures that the peak point has a significant high-level characteristic and avoids misjudging ordinary high values ​​as peak points.

[0081] The fallback point is defined as the first node after the peak value where the regulation capacity constraint value falls below 100% of the mean. The continuous period from the trough point to the fallback point constitutes a complete fluctuation cycle. For example, if the regulation capacity constraint values ​​for a certain period within a scheduling cycle are 30 MW, 45 MW, 60 MW, 85 MW, 90 MW, 88 MW, 75 MW, 60 MW, and 40 MW respectively, with the trough point at 30 MW, the peak point at 90 MW, and the fallback point at 60 MW, then this fluctuation cycle is the corresponding continuous period.

[0082] In each fluctuation cycle, the continuous period when the regulation capacity is at a high level is determined by a threshold. In this embodiment, the high level threshold is set to 80% of the maximum value of the regulation capacity constraint within the fluctuation cycle. This threshold is determined based on the effective utilization range of the grid absorption capacity. The 80% ratio can ensure that the selected period has sufficient additional absorption capacity, and can also avoid the problem of too few candidate windows due to the threshold being too high, or insufficient window absorption capacity due to the threshold being too low.

[0083] It should be added that if no continuous period reaches the high threshold within a certain fluctuation cycle, or if the length of the continuous period that reaches the threshold is less than 2 hours, then the fluctuation cycle is discarded and no candidate window is generated. The minimum duration of 2 hours is determined based on the actual feasibility of the scheduling operation. A window that is too short cannot complete the concentrated release of the non-followable fluctuation components and has no practical scheduling value.

[0084] Furthermore, the selection of candidate windows needs to consider both the time span and energy magnitude of the unfollowable fluctuation components. The time span of the unfollowable fluctuation component is the number of consecutive time nodes corresponding to its non-zero power values. The energy magnitude is the cumulative energy of the unfollowable fluctuation component, which is the sum of each non-zero power value multiplied by the time granularity.

[0085] During the screening process, the candidate window duration must be no less than 1.2 times the time span of the non-followable fluctuation component. The 1.2 times coefficient is a reserved scheduling buffer time to ensure that the fluctuation component can smoothly enter the window and avoid power surges caused by time constraints.

[0086] Secondly, the total absorption capacity of the candidate window is required to be no less than the energy of the non-followable fluctuation component. The total absorption capacity is calculated as the sum of the differences between the adjustment capacity constraint values ​​of all time nodes within the window and the followable fluctuation component of that node.

[0087] In another implementation of this step, a dynamic threshold adjustment mechanism can be introduced. Based on the statistical distribution characteristics of the regulation capacity constraints within future scheduling cycles, the mean and standard deviation of the regulation capacity constraints are calculated. The high threshold for each fluctuation cycle is set as the mean plus 1.2 times the standard deviation. The coefficient of 1.2 times is determined based on the probability characteristics of the normal distribution, which can ensure that the high threshold covers the high probability range of regulation capacity, while adapting to the changes in regulation capacity in different scheduling cycles.

[0088] It should be added that other constraints on power grid operation must be considered during the window selection process, such as transmission channel capacity limitations and regional load balancing requirements. If the transmission channel capacity is close to its upper limit, the load rate exceeds 95%, or the regional load gap is small, less than 3% of the total load, during the time period corresponding to a candidate capacity relaxation window, then the priority of that window should be reduced, or even abandoned, to ensure the practicality and safety of the capacity relaxation window.

[0089] The curve generation and translation device is used to separate the non-followable fluctuation component from the original time period and translate it as a whole into the capacity relaxation window for concentrated release, so as to generate a target power generation curve. The fluctuation pattern of the target power generation curve matches the spatiotemporal distribution of the grid's dynamic adjustment capability.

[0090] The curve generation and translation device is used to translate the unfollowable fluctuation component as a whole on the time axis according to the start time and duration of the capacity relaxation window, so that it falls completely within the capacity relaxation window, and to align the peak time of the unfollowable fluctuation component with the peak time of the adjusted capacity within the capacity relaxation window.

[0091] The curve generation and translation device is also used to reshape the waveform of the non-followable fluctuation component after translating the non-followable fluctuation component into the capacity relaxation window: to obtain the adjustment rate constraint change curve within the capacity relaxation window; Based on the adjustment rate constraint change curve, the internal morphology of the shifted non-followable fluctuation component is adjusted so that the instantaneous rate of change of the adjusted non-followable fluctuation component does not exceed the adjustment rate constraint value at any time within the capacity relaxation window, while keeping the total energy of the non-followable fluctuation component constant.

[0092] When adjusting the internal shape of the non-followable fluctuation component after translation, the curve generation and translation device is used to: identify the original waveform characteristics of the non-followable fluctuation component, the original waveform characteristics including: Rise steepness: The average rate of change (megawatts per minute) of power within a sub-segment from the lowest point to the highest point. Peak amplitude: The highest power value within a sub-segment; Falling slope: The average rate of change of power within a sub-segment from its highest point to its lowest point; Based on the changing trend (from high to low) of the adjustment rate constraint within the capacity relaxation window, the unfollowable fluctuation component is divided into multiple sub-segments. The division rules are as follows: Calculate the power change rate difference between adjacent time nodes in the original waveform. If the absolute value of the difference exceeds 20% of the average adjustment rate constraint of the current scheduling cycle, then the node is taken as the dividing point between different sub-segments. If the change rate difference of multiple consecutive nodes is continuously less than the threshold, then they are merged into one sub-segment to ensure that each sub-segment has similar steepness characteristics.

[0093] The multiple sub-segments are reordered and spliced ​​within the capacity relaxation window. The sorting rule is as follows: the sub-segments are arranged in descending order of the steepness of the rising segment, corresponding to the time periods within the capacity relaxation window from high to low adjustment rate constraints. If the number of sub-segments is greater than the number of time periods that can be allocated within the window, the multiple sub-segments with the smallest steepness are merged into a whole and still placed in the time period with the lowest constraint.

[0094] During splicing, the endpoint power of adjacent sub-segments is smoothly transitioned through linear interpolation to ensure the overall continuity of the waveform. During the adjustment process, the power values ​​within each sub-segment are adjusted (scaled proportionally while keeping the total energy of the sub-segments constant) so that the total energy of the re-spliced ​​waveform is equal to the total energy of the original non-followable wave components.

[0095] Specifically, the start time of the capacity relaxation window is the moment corresponding to the first time node of the window, and the duration is the number of time nodes within the window multiplied by the time granularity.

[0096] The translation of the unfollowable fluctuation component first determines the peak time within its original time span. The peak time is defined as the time corresponding to the time node with the largest power value in the unfollowable fluctuation sequence.

[0097] Then, the peak time of the adjustment capacity within the capacity relaxation window is calculated, which is the time corresponding to the time node with the largest adjustment capacity constraint value within the window.

[0098] During the shift, the two peak times are perfectly aligned by adjusting the time axis offset. This alignment strategy is determined based on the principle of maximizing the utilization of the grid's peak absorption capacity. The peak time of the regulating capacity is the period when the grid's absorption capacity is strongest. Aligning with the peak time of the non-followable fluctuation component ensures that the core energy of the fluctuation component is effectively absorbed, while avoiding insufficient absorption capacity due to peak mismatch. At the same time, it ensures that all non-zero power nodes of the non-followable fluctuation component fall within the time range of the capacity relaxation window.

[0099] The regulation rate constraint change curve within the capacity relaxation window is obtained through the real-time data interface of the power grid dispatch center. This curve is continuous time-series data with a time resolution consistent with the dispatch cycle, covering the regulation rate constraint values ​​of all time nodes within the window. The curve shape reflects the dynamic change characteristics of the power grid regulation rate within the window, and this change curve is the core basis for waveform reshaping.

[0100] During the acquisition process, the system will perform integrity verification on the curve data. If there are missing data nodes, cubic spline interpolation will be used to complete them, ensuring the continuity and accuracy of the curve. The cubic spline interpolation method is chosen based on its characteristics of good smoothness and small error in time series data completion, which can restore the true change trend of the adjustment rate constraint to the greatest extent.

[0101] The waveform reshaping process first identifies the waveform features of the non-followable wave components after translation, and extracts the rising slope, peak amplitude, and falling slope respectively. The rising slope represents the instantaneous rate of change during the power increase phase, the falling slope represents the instantaneous rate of change during the power decrease phase, and the peak amplitude represents the maximum output level of the wave component.

[0102] After feature recognition is completed, the non-followable fluctuation component is divided into several sub-segments with similar power change characteristics according to the changing trend of the adjustment rate constraint from high to low within the capacity relaxation window. Each sub-segment has independent steepness characteristics and energy proportion.

[0103] Subsequently, the sub-segments were reordered and spliced ​​according to the principle of matching steepness and constraints. Sub-segments with steeper rising segments were placed in periods with higher regulation rate constraints, and sub-segments with steeper falling segments were placed in periods with the second highest regulation rate constraints. The remaining sub-segments were arranged in periods with relatively lower constraints. During the splicing process, the waveform was kept continuous by smoothly connecting the power values ​​at the endpoints of adjacent sub-segments, and the total energy was conserved by keeping the energy ratio of each sub-segment constant.

[0104] The core logic of waveform reshaping is rate constraint adaptation plus energy conservation. First, the instantaneous rate of change of the non-followable fluctuation component after translation is calculated at each time node within the window. The instantaneous rate of change is calculated by dividing the power difference between the current node and the previous node by the time interval. The instantaneous rate of change is compared with the corresponding adjustment rate constraint value at each point. If the instantaneous rate of change of a certain node exceeds the constraint value, the shape adjustment process is initiated.

[0105] Waveform reshaping employs an iterative correction method to ensure that the shifted, non-followable fluctuation components satisfy the regulation rate constraint within the capacity relaxation window, while maintaining a constant total energy. The specific steps are as follows: (1) Point-by-point check: Calculate the power change rate of each time node relative to the previous node within the calculation window. If the change rate of a node exceeds the adjustment rate constraint at that moment, the power of that node is forcibly adjusted to the power of the previous node plus (or minus) the maximum change value allowed by the constraint, so that the change rate of that node is compliant.

[0106] (2) Energy Compensation: Calculate the total energy change ΔE (megawatt-hours) caused by the forced adjustment. Let Psum (megawatts) be the sum of the power values ​​of all non-violation nodes within the window. Then, the power increment to be compensated for each non-violation node i is (ΔE / time granularity) × (Pi / Psum), where the time granularity is 0.25 hours. Add the corresponding compensation value to each non-violation node to restore the total energy of the window to its original value.

[0107] (3) Iteration loop: Repeat steps (1) and (2) until all node change rates meet the constraints and the maximum power change in the two iterations is less than the preset threshold (e.g., 0.1 MW). If the iteration fails to converge after more than 50 iterations, the node values ​​are fine-tuned according to the power ratio and the process is forcibly terminated.

[0108] After waveform reshaping is completed, the system will generate a power sequence of the reshaped non-followable fluctuation components and compare it with the original sequence for verification. The verification dimensions include total energy deviation, which must be controlled within ±0.1%, compliance of instantaneous change rate of each node, and peak moment matching degree. The peak moment after reshaping is still aligned with the peak moment of the window regulation capacity. This verification standard is determined based on the core objectives of waveform reshaping to ensure that the adjusted fluctuation components not only comply with the grid regulation rate constraints, but also do not change the core energy characteristics and peak utilization strategy.

[0109] If the verification fails, the iterative optimization process is restarted, and the allocated weights and number of iterations are adjusted until all verification requirements are met.

[0110] Within the capacity relaxation window, the new energy power plant is controlled to generate electricity at a rate exceeding the regulation rate constraint, so as to utilize the peak absorption capacity of the power grid during that period to absorb the unfollowable fluctuation component.

[0111] Outside the capacity relaxation window, the output curve of the new energy power plant is controlled to remain consistent with the followable fluctuation component, so that its output speed is always limited by the adjustment rate constraint.

[0112] Understandably, the power grid has a stronger ability to absorb fluctuations within the capacity relaxation window, thus allowing power plants to exceed the constraints of the conventional regulation rate. The output rate exceeding the regulation rate constraint is set to 2 to 3 times the conventional regulation rate constraint. In this embodiment, it is set to 3 times, that is, the maximum output change rate does not exceed 15 MW / min, and the conventional constraint upper limit of 5 MW / min is multiplied by 3. This multiple is determined based on the short-circuit capacity and frequency response characteristics of the power grid. Through extensive simulation verification, 3 times the conventional rate can not only achieve rapid absorption of non-followable fluctuation components, but also ensure that the power grid frequency fluctuation is controlled within a safe range of ±0.1 Hz, without causing power grid stability problems.

[0113] The output speed is controlled by increasing and decreasing in stages. Starting from the beginning of the window, the power generation is increased at a rate of 10 MW per minute. The stage increase rate of 10 MW per minute is determined based on the actual adjustment and response capability of the power plant's power generation unit to avoid instantaneous power surges from impacting the power grid.

[0114] Meanwhile, the system monitors the grid frequency response in real time. If the frequency deviates from 50 Hz plus or minus 0.1 Hz, the output adjustment rate is immediately reduced to 5 MW per minute to ensure the safe operation of the grid.

[0115] Outside of the capacity relaxation window, the grid's regulation capacity is relatively limited, and output must be strictly controlled in accordance with regulation rate constraints.

[0116] Consistency control between the output curve and the followable fluctuation component is achieved through node-by-node power tracking. The absolute value of the deviation between the actual output value and the power value corresponding to the followable fluctuation component at each time node does not exceed 1 MW. This deviation threshold is determined based on the power tracking accuracy requirements of the power grid dispatch, which can ensure the stability and controllability of the output curve.

[0117] The output speed is limited by calculating the power change rate of adjacent nodes in real time. If the calculated change rate exceeds the adjustment rate constraint at the corresponding moment, the output value of the current node is automatically adjusted to reduce the change rate to within the constraint range.

[0118] In this embodiment, the process of stripping the unfollowable fluctuation component corresponds to the fluctuation decomposition process. The power value of the unfollowable fluctuation component is extracted node by node in chronological order to form an unfollowable fluctuation sequence. The time length of this sequence is consistent with the original scheduling cycle, but non-zero power values ​​exist only at some time nodes.

[0119] When selecting a target capacity relaxation window, the system prioritizes the candidate window with the highest priority. The total absorption capacity of the capacity relaxation window is the sum of the differences between the adjusted capacity constraint values ​​at all time points within the window and the fluctuation components that the node can follow.

[0120] If the total absorption capacity of the window can fully accommodate the accumulated energy of the unfollowable fluctuation components, then the entire unfollowable fluctuation sequence will be shifted into the window.

[0121] The translation of non-followable fluctuation components must follow the principle of power distribution uniformity. The total energy of the non-followable fluctuation sequence should be evenly distributed to each time node within the window according to the window duration to ensure that the power change rate after translation does not exceed the adjustment rate constraint at the corresponding time of the window and the maximum allowable output change rate within the window.

[0122] The target power generation curve is generated by superimposing the power sequence of the followable fluctuation component with the shifted power sequence of the non-followable fluctuation component. Outside the capacity relaxation window during the original time period, only the power value of the followable fluctuation component is retained. During the capacity relaxation window, the power values ​​of the followable fluctuation component and the shifted power values ​​of the non-followable fluctuation component are superimposed to form the complete power sequence of the target power generation curve.

[0123] After generation, the target power generation curve needs to be verified to ensure that the power value at each time node does not exceed the regulation capacity constraint at the corresponding time, and that the power change rate of nodes outside the window does not exceed the conventional regulation rate constraint, and the power change rate of nodes inside the window does not exceed the maximum allowable output change rate, which is 3 times the conventional constraint.

[0124] If any constraints are not met, the power value of that node is fine-tuned. The fine-tuning employs gradient descent, chosen because it offers advantages in constrained optimization problems, such as fast convergence and high adjustment accuracy, enabling rapid correction of the power value to within the constraint range.

[0125] A verification device is used to perform dynamic compatibility verification on the target power generation curve after the curve generation and translation device translates the non-followable fluctuation component to the capacity relaxation window.

[0126] The dynamic compatibility verification includes: verifying whether the maximum output speed of the shifted non-followable fluctuation component within the capacity relaxation window exceeds the allowable upper limit of the regulation rate constraint for that period, and verifying whether the cumulative power generation of the target power generation curve at any time exceeds the regulation capacity constraint for the corresponding time.

[0127] If the verification passes, the target power generation curve is output; if the verification fails, the translation position is adjusted or the backup absorption mechanism is activated.

[0128] Specifically, dynamic compatibility verification is a key closed-loop process to ensure that the final target power generation curve conforms to the safe operation boundary of the power grid.

[0129] The verification process is triggered after the curve is generated but before the command is issued. It is carried out by combining time-by-time node traversal with global energy statistics to ensure the comprehensiveness and real-time nature of the verification.

[0130] The verification device reads the power sequence of the target power generation curve, the allowable upper limit parameter of the capacity relaxation window, and the real-time adjustment capacity constraint data of the power grid, and then executes the subsequent dual verification logic.

[0131] The first verification is the rate constraint compliance verification. First, the power sequence of the non-followable fluctuation component after translation within the capacity relaxation window is extracted. The power difference between adjacent time nodes in the sequence is calculated and divided by the time interval to obtain the actual output rate of each node within the window. Then, the actual output rate of each node is compared with the upper limit of the regulation rate constraint at the corresponding time within the window. To adapt to the dynamic fluctuations of the power grid, the upper limit is set to 1.5 to 2 times the conventional regulation rate constraint. If the actual output rate of any node exceeds the upper limit, the rate verification is deemed to have failed.

[0132] The second verification is the capacity constraint compliance verification. The cumulative power generation of the target power generation curve at any time is calculated. This cumulative value is the sum of the power values ​​and time granularity products of all time nodes from the start of the scheduling cycle to the current time. At the same time, the regulation capacity constraint value of the power grid at the corresponding time is queried. If the cumulative power generation exceeds the regulation capacity constraint value, the capacity verification is deemed to have failed. Both verifications must be satisfied simultaneously for the overall verification to be considered as passed.

[0133] If the verification fails, the system will trigger a differentiated processing mechanism based on the specific type of failure. If the reason for the failure is that the maximum output speed within the window exceeds the allowable upper limit, the verification device will send a translation position adjustment command to the curve generation and translation device. The adjustment strategy prioritizes power smoothing reconstruction within the window. That is, while keeping the total energy of the non-followable fluctuation component unchanged, the power values ​​of each time node within the window are redistributed through linear interpolation or spline interpolation algorithms to make the power change curve smoother and ensure that the adjusted output speed is strictly lower than the allowable upper limit. If a single window cannot meet the requirements through power reconstruction, a window switch is triggered. The system automatically searches for the next lower priority capacity relaxation window, attempts to translate the non-followable fluctuation component to the new window, and re-performs the compatibility verification.

[0134] If the verification fails because the cumulative power generation exceeds the regulation capacity constraint, the system will directly activate the backup absorption mechanism. The backup absorption mechanism is taken over by the collaborative absorption device, which will allocate the energy gap exceeding the regulation capacity constraint to the backup charging and discharging modules of the energy storage system or the emergency load pool of the demand-side response resources. Specific instructions include instructing the energy storage system to charge and absorb at a rate of not less than 4 MW per minute, or instructing the demand-side response resources to reduce the instantaneous load by not less than 5 MW, until the cumulative power generation of the target power generation curve fully meets the regulation capacity constraint requirements.

[0135] If the target power generation curve still fails the compatibility check after the above adjustments, the system will determine that the absorption capacity is insufficient in the current dispatch cycle, trigger emergency power curtailment protection and early warning, send alarm information to the power grid dispatch center and the new energy power station control center, and rigidly reduce the non-followable fluctuation components according to the preset minimum power curtailment strategy to ensure the safe operation of the main power grid.

[0136] A collaborative absorption device is used to calculate the energy gap between the total energy of the non-followable fluctuation component and the remaining absorption capacity of the capacity relaxation window when there is no capacity relaxation window that can fully accommodate the non-followable fluctuation component within the future scheduling period.

[0137] The energy gap is allocated to energy storage systems and demand-side response resources to generate coordinated consumption instructions.

[0138] According to the coordinated absorption command, the energy storage system is activated for charging and discharging, and the demand-side load is adjusted to assist in the absorption of the remaining portion of the non-followable fluctuation component.

[0139] Specifically, the remaining absorption capacity of all identified capacity relaxation windows is first calculated. The remaining absorption capacity is defined as the remaining value after subtracting the allocated energy of the non-followable fluctuation component from the total absorption capacity of each capacity relaxation window. If there is no capacity relaxation window, the total remaining absorption capacity is 0.

[0140] The total energy of the non-followable fluctuation component is its cumulative energy over the entire scheduling cycle, calculated by multiplying each non-zero power value by the time granularity. The energy gap is calculated by subtracting the sum of the remaining absorption capacity of all capacity relaxation windows from the total energy of the non-followable fluctuation component. When the calculation result is positive, an energy gap is determined to exist. This calculation logic is based on the principles of energy conservation and absorption balance, ensuring accurate quantification of the energy scale that needs to be absorbed through coordinated resources.

[0141] The energy gap allocation adopts a strategy of prioritizing energy storage and supplementing demand-side resources. This strategy is determined based on the characteristics of energy storage systems, such as fast response speed and high adjustment accuracy, prioritizing the rapid mitigation role of energy storage. The allocation ratio is set at 60% to 80% for energy storage systems and 20% to 40% for demand-side response resources. In this embodiment, it is set at 70% for energy storage systems and 30% for demand-side resources. This ratio is determined based on the current actual adjustment capabilities of energy storage and demand-side resources in my country, avoiding overloading of energy storage and resulting in lifespan loss, while fully mobilizing the flexibility of demand-side resources. The coordinated consumption command includes parameters such as the energy to be consumed, the response period, and the adjustment rate. The response period prioritizes matching the original period that cannot follow fluctuation components or the adjacent period of the capacity relaxation window to ensure the rationality of the consumption sequence. The adjustment rate requires energy storage systems to be no less than 3 MW / min and demand-side response resources to be no less than 1 MW / min. This rate requirement is determined based on the real-time requirements of coordinated consumption to quickly absorb the energy gap.

[0142] The charging and discharging control of the energy storage system is executed based on the energy gap and response time period in the coordinated absorption command. When the non-followable fluctuation component is power excess, the energy storage system starts the charging mode and absorbs the excess energy according to the adjustment rate required by the command. The charging power does not exceed 80% of the rated power of the energy storage system. This limit is determined based on the safe charging and discharging threshold of the energy storage system to avoid equipment damage caused by overcharging.

[0143] When the non-followable fluctuation component indicates insufficient power, the energy storage system initiates a discharge mode to release stored energy to fill the gap, with the discharge power not exceeding 80% of the rated power. Load adjustment of demand-side response resources is achieved through commands issued by the aggregator platform, prioritizing shiftable loads such as electric vehicle charging and flexible industrial loads, and interruptible loads such as commercial air conditioning and non-core production loads. Adjustments must ensure the reliability of power supply to the user side, with the cumulative interruption time not exceeding one hour per day. This time limit is determined based on a balance between user-side power consumption experience and grid dispatching needs, avoiding excessive impact on normal user power consumption.

[0144] Meanwhile, the system monitors the state of charge of the energy storage system and the actual adjustment of demand-side response resources in real time. If the state of charge of the energy storage is below 20%, it will automatically transfer the unfinished consumption tasks to the demand-side response resources. If the actual adjustment of the demand-side response resources is insufficient, it will feed back to the system and activate the backup demand-side resources to ensure that the energy gap is completely absorbed.

[0145] A control actuator is used to control the power generation unit of the new energy power station according to the target power generation curve.

[0146] Specifically, the control and execution device adopts a distributed control architecture. This architecture is chosen based on the characteristics of the dispersed arrangement of power generation units in new energy power plants. The distributed architecture can reduce the communication delay and single point of failure risk of centralized control and improve control reliability.

[0147] Communication connections are established with the controllers of each power generation unit in the new energy power plant via industrial Ethernet, with communication latency controlled within 100 milliseconds. This latency threshold is determined based on the real-time requirements of power generation control. 100 milliseconds ensures that the target power command is issued in a timely manner, avoiding output deviations caused by command delays. The power sequence of the target power generation curve is issued to each power generation unit controller according to time nodes, with the issuance frequency consistent with the time granularity, i.e., the target power command for that period is issued once every 15 minutes.

[0148] For photovoltaic power plants, the power generation unit control adopts a combination of maximum power point tracking regulation and power limiting. Based on the target power command of the target power generation curve, the output power of the photovoltaic inverter is adjusted. Power regulation is achieved by controlling the modulation ratio of the inverter, with the adjustment accuracy controlled within ±1%. This accuracy is determined according to the industry standards for power grid control of new energy power plants and can meet dispatch assessment requirements.

[0149] If the target power command is lower than the current maximum available power, the inverter will operate in power-limiting mode, reducing the output power to the target value.

[0150] If the target power command is higher than the current maximum available power, the inverter operates in maximum power point tracking mode, outputs the maximum available power, and feeds back power shortage information to the system.

[0151] Within the capacity relaxation window, the inverter's modulation ratio adjustment rate is increased to twice that of the conventional mode. This increase is determined based on the inverter's hardware adjustment capability, enabling it to quickly respond to power changes within the window. At the same time, it absorbs power fluctuations through the capacitor energy storage module to maintain stable output voltage.

[0152] For wind power plants, the power generation unit control is achieved by adjusting the pitch angle and speed of the wind turbine. Based on the target power command of the target power generation curve, the wind turbine controller calculates the optimal pitch angle and speed values, adjusts the pitch angle through the hydraulic system, and controls the generator speed through the converter to make the wind turbine output power reach the target value.

[0153] During the adjustment process, the pitch angle adjustment rate should not exceed 0.5 degrees per second and the speed adjustment rate should not exceed 5 revolutions per minute during normal periods. This rate limit is determined based on the fatigue life requirements of the wind turbine's mechanical structure to avoid excessively rapid adjustments that could lead to equipment damage.

[0154] Within the capacity relaxation window, the pitch angle adjustment rate is increased to 1 degree per second, and the speed adjustment rate is increased to 8 revolutions per minute, in order to achieve output control exceeding the conventional adjustment rate. This increased rate has been verified by the wind turbine manufacturer for safety and will not affect equipment safety in short-term operation.

[0155] Meanwhile, the wind turbine's load monitoring system monitors the stress on the blades and tower in real time. If the load exceeds 90% of the design threshold, the adjustment rate will be automatically reduced to ensure equipment safety.

[0156] Using the above method, the control actuator will monitor the actual output power of each power generation unit in real time, collect output current and voltage data through current transformers and voltage transformers, calculate the actual output power, and the sampling frequency is once every 1 second. This sampling frequency is determined according to the accuracy requirements of power monitoring. Once every 1 second can capture power fluctuations in time and provide data support for closed-loop regulation.

[0157] The actual output power is compared with the target power command. If the absolute value of the deviation exceeds 2%, closed-loop regulation is initiated, and the control parameters are adjusted until the deviation is controlled within the allowable range. The 2% deviation threshold is determined according to the power grid dispatch assessment standards and can meet the requirements of conventional dispatch.

[0158] In addition, the control and execution device also has an emergency response function. The priority of the emergency power adjustment command is divided into three levels: Level 1 is the abnormal grid frequency, with a frequency deviation of 50 Hz ± 0.2 Hz; Level 2 is the overload of the transmission channel, with a load rate exceeding 95%; and Level 3 is the regional load gap exceeding 5%. This priority division is determined according to the importance of the safe operation of the power grid. Frequency abnormalities directly affect the stability of the power grid, so they have the highest priority.

[0159] After receiving an emergency command, the system first determines the command priority. If it is a level one command, the target power generation curve is immediately cut off and the power generation is adjusted to the command-required value within 1 second. The 1-second response time is determined according to the technical standards for emergency control of power grid frequency.

[0160] If it is a Level 2 or Level 3 command, the adjustment will be completed within 3 seconds. The 3-second response time can balance the adjustment speed and equipment safety. During the adjustment process, the power change rate is ensured not to exceed 5 megawatts per minute. The upper limit of the adjustment rate in emergency situations is determined based on the maximum capacity of the power grid in emergency situations.

[0161] If the power grid dispatch center issues an emergency power adjustment command, or detects a fault in a power generation unit, the system will prioritize responding to the emergency command, suspend the execution of the target power generation curve, adjust the power generation according to the emergency command, and resume the normal execution of the target power generation curve after the fault is cleared or the emergency is resolved.

[0162] The beneficial effects of this invention are as follows: First, this invention, through a fluctuation decomposition device, decomposes the predicted power generation curve into followable and non-followable fluctuation components based on the regulation rate constraint, achieving refined analysis of the power generation curve. Specifically, by splitting continuous fluctuations into independent fluctuation events and determining whether each event can be accommodated, or by comparing the predicted curve with the regulation rate constraint point by point and marking fluctuation segments exceeding the constraint, it can accurately and completely extract fluctuation components that the grid regulation rate cannot follow. This provides a clear decomposition basis for subsequent optimized scheduling and avoids the problem of missed or misjudged fluctuation components caused by the overall curve processing in traditional schemes.

[0163] Secondly, this invention, through a window recognition device, determines the capacity relaxation window based on the time-varying characteristics of the regulation capacity constraint, thereby achieving precise spatiotemporal positioning of the grid's absorption capacity. Specifically, by predicting the trend of regulation capacity changes, identifying complete fluctuation cycles, and screening windows that can fully accommodate non-followable fluctuation components, it can accurately pinpoint continuous periods when the grid has additional absorption capacity, maximizing the utilization of the grid's spatiotemporal regulation potential. This allows fluctuation components that would otherwise be reduced due to insufficient capacity to have a concentrated absorption opportunity, significantly increasing the space for renewable energy absorption.

[0164] Third, this invention, through a curve generation and translation device, shifts the unfollowable fluctuation component as a whole into a capacity relaxation window for concentrated release and reshapes its waveform, achieving a deep match between the power generation curve and the grid's regulation capability. Specifically, by aligning the peak time of the unfollowable fluctuation component with the peak time of the regulation capacity within the window, it ensures that the core energy is absorbed when the grid's absorption capacity is at its strongest. By identifying the original waveform characteristics, dividing it into multiple sub-segments, and reordering and splicing them according to the trend of regulation rate constraint changes, segments with high steepness are matched with periods of high constraint. Under the premise of keeping the total energy constant, the instantaneous rate of change of the adjusted fluctuation component at any moment within the window does not exceed the corresponding constraint value. This mechanism fundamentally avoids the problem of instantaneous rate of change exceeding the limit, making the fluctuation pattern of the target power generation curve highly compatible with the spatiotemporal distribution of the grid's dynamic regulation capability, and reducing the secondary regulation burden of the grid.

[0165] Fourth, this invention achieves a closed-loop guarantee of safety and feasibility by dynamically verifying the target power generation curve through a verification device. Specifically, by verifying whether the maximum output speed within the verification window exceeds the upper limit allowed by the regulation rate constraint, and whether the cumulative power generation at any time exceeds the regulation capacity constraint, it ensures that the final output target power generation curve strictly conforms to the grid's safe operation boundary. If the verification fails, a position adjustment or backup absorption mechanism is triggered, fundamentally ensuring the high compatibility between the target power generation curve and grid operation, and avoiding grid stability risks caused by improper dispatch instructions.

[0166] Fifth, this invention, through a collaborative absorption device, allocates the energy gap to energy storage systems and demand-side response resources when there is no capacity relaxation window that can fully accommodate unresponsive fluctuation components, thus achieving the collaborative participation of multiple types of resources. Specifically, by calculating the energy gap and adopting an allocation strategy that prioritizes energy storage and supplements demand-side resources, a collaborative absorption command is generated and energy storage charging and discharging and demand-side load adjustment are initiated. This can completely solve the energy gap problem when the grid's own absorption capacity is insufficient, avoid the waste of new energy generation potential caused by the rigid reduction of traditional solutions, and fully mobilize the potential of rapid response from energy storage and flexible adjustment from the demand side.

[0167] Sixth, this invention achieves a unification of theoretical optimization and practical execution by controlling the power generation units of a new energy power plant precisely according to the target power generation curve through a control execution device. Specifically, it adopts a distributed control architecture, employing adapted control strategies for photovoltaic and wind power plants respectively. Real-time monitoring and closed-loop regulation ensure that the deviation between actual output and target commands is controlled within allowable limits. It also possesses an emergency response function, prioritizing grid safety in emergency situations. This mechanism ensures the accurate implementation of the target power generation curve, improving the reliability and precision of power plant power generation control.

[0168] In summary, this invention can fully tap the potential of new energy power generation and significantly improve the efficiency of new energy consumption while ensuring the safe and stable operation of the power grid. At the same time, it can achieve the lowest comprehensive regulation cost through optimized calculation of regulation costs.

[0169] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A power generation curve management system for new energy power plants based on intelligent power dispatching, characterized in that, include: The information acquisition module is used to acquire dynamic regulation capability information of the power grid in the future dispatch cycle, and to acquire the predicted power generation curve of the new energy power station in the future dispatch cycle; the dynamic regulation capability information includes regulation rate constraints and regulation capacity constraints that change over time. The fluctuation decomposition module is used to decompose the predicted power generation curve into a followable fluctuation component and a non-followable fluctuation component according to the adjustment rate constraint; wherein, the followable fluctuation component refers to the fluctuation part whose rate of change never exceeds the adjustment rate constraint at the corresponding time, and the non-followable fluctuation component refers to the fluctuation part whose rate of change exceeds the adjustment rate constraint at the corresponding time. The window identification module is used to determine at least one capacity relaxation window within the future scheduling cycle based on the time-varying characteristics of the adjustment capacity constraint; the capacity relaxation window refers to a continuous period of time when the adjustment capacity constraint value is high and the power grid has additional absorption capacity. The curve generation and translation module is used to separate the unfollowable fluctuation component from the original time period and translate it as a whole into the capacity relaxation window for centralized release, so as to generate a target power generation curve. The fluctuation pattern of the target power generation curve matches the spatiotemporal distribution of the grid's dynamic adjustment capability. The control execution module is used to control the power generation unit of the new energy power station according to the target power generation curve.

2. The power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 1, characterized in that, The window recognition module is used for: Based on the prediction of future regulation capacity change trends, the curve of the regulation capacity constraint changing over time is analyzed to identify the complete fluctuation cycle of regulation capacity from low to high and then back down. In each fluctuation cycle, the continuous period when the regulation capacity is at a high level is identified as the candidate window; Based on the time span and energy magnitude of the unfollowable fluctuation component, a window that can fully accommodate the unfollowable fluctuation component is selected from the candidate windows and used as the capacity relaxation window.

3. The power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 2, characterized in that, The curve generation and translation module is used for: Based on the start time and duration of the capacity relaxation window, the unfollowable fluctuation component is shifted as a whole on the time axis so that it falls completely within the capacity relaxation window, and the peak time of the unfollowable fluctuation component is aligned with the peak time of the adjusted capacity within the capacity relaxation window. Within the capacity relaxation window, the new energy power plant is controlled to generate electricity at a rate exceeding the regulation rate constraint, so as to utilize the peak absorption capacity of the power grid during this period to absorb the unfollowable fluctuation component. Outside the capacity relaxation window, the output curve of the new energy power plant is controlled to remain consistent with the followable fluctuation component, so that its output speed is always limited by the adjustment rate constraint.

4. The power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 1, characterized in that, The wave decomposition module is used for: The predicted power generation curve is compared point by point with the regulation rate constraint; The fluctuation segments whose rate of change exceeds the adjustment rate constraint are marked, and the fluctuation segments are spliced ​​together to form the unfollowable fluctuation component; The followable fluctuation component is obtained by subtracting the non-followable fluctuation component from the predicted power generation curve.

5. The power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 1, characterized in that, Before decomposition, the wave decomposition module is also used for: The continuous fluctuations in the predicted power generation curve are divided into several independent fluctuation events, each with a clear start point, peak point, and end point; Each fluctuation event is determined to be fully accommodated by the adjustment rate constraint. Fluctuations that cannot be accommodated are marked as the unfollowable fluctuation components.

6. A power generation curve management system for new energy power plants based on intelligent power dispatching as described in claim 1, characterized in that, It also includes a coordinated absorption module, used when there is no capacity relaxation window within the future scheduling period that can fully accommodate the unfollowable fluctuation component: Calculate the energy gap between the total energy of the non-followable fluctuation component and the remaining absorption capacity of the capacity relaxation window; The energy gap is allocated to energy storage systems and demand-side response resources to generate coordinated consumption instructions; According to the coordinated absorption command, the energy storage system is activated for charging and discharging, and the demand-side load is adjusted to assist in the absorption of the remaining portion of the non-followable fluctuation component.

7. A power generation curve management system for new energy power plants based on intelligent power dispatching as described in claim 1, characterized in that, It also includes a verification module, used after the curve generation and translation module translates the unfollowable fluctuation component to the capacity relaxation window: The target power generation curve is dynamically compatibility verified; The dynamic compatibility verification includes: verifying whether the maximum output speed of the shifted non-followable fluctuation component within the capacity relaxation window exceeds the allowable upper limit of the regulation rate constraint for that period, and verifying whether the cumulative power generation of the target power generation curve at any time exceeds the regulation capacity constraint for the corresponding time. If the verification passes, the target power generation curve is output; if the verification fails, the translation position is adjusted or the backup absorption mechanism is activated.

8. A power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 1, characterized in that, The dynamic adjustment capability information includes the adjustment response speed, adjustment duration, and adjustment cost of different types of power supplies; The regulation rate constraint and the regulation capacity constraint are generated through optimization calculations based on the regulation response speed, regulation duration, and regulation cost, so that the generated constraints minimize the overall regulation cost while ensuring power grid security.

9. A power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 3, characterized in that, The curve generation and translation module is also used to reshape the waveform of the unfollowable fluctuation component after translating it into the capacity relaxation window: Obtain the adjustment rate constraint change curve within the capacity relaxation window; Based on the adjustment rate constraint change curve, the internal morphology of the shifted non-followable fluctuation component is adjusted so that the instantaneous rate of change of the adjusted non-followable fluctuation component does not exceed the adjustment rate constraint value at any time within the capacity relaxation window, while keeping the total energy of the non-followable fluctuation component constant.

10. A power generation curve management system for new energy power plants based on intelligent power dispatching according to claim 9, characterized in that, The curve generation and translation module, when adjusting the internal shape of the translated non-followable fluctuation component, is used for: Identify the original waveform characteristics of the unfollowable fluctuation component, the original waveform characteristics including the steepness of the rising segment, the peak amplitude, and the steepness of the falling segment; Based on the changing trend of the adjustment rate constraint within the capacity relaxation window, the unfollowable fluctuation component is divided into multiple sub-segments; The multiple sub-segments are reordered and spliced ​​within the capacity relaxation window, so that the sub-segments with a steeper rising segment correspond to the time period with a higher adjustment rate constraint, and the sub-segments with a steeper falling segment correspond to the time period with the second highest adjustment rate constraint, and the re-spliced ​​waveform remains continuous and the total energy remains unchanged.

Images