A clean energy base multi-time scale error compensation backup scheduling method

Abstract

A clean energy base multi-time scale error compensation standby scheduling method, comprising: making multi-time scale prediction of wind, light, water and load in day-ahead, day-ahead and real-time, comparing the predicted value with the historical actual value to obtain the predicted absolute error distribution, determining the error compensation coefficient of each time scale and each type of energy according to the probability lower limit of preventing extreme prediction, and taking the product of the error compensation coefficient and the corresponding predicted value as the error compensation demand; taking the lower bound of the error compensation demand, the maximum climbing rate term of hydropower and the upper limit term of load of each cascade hydropower station at each time as the error compensation standby capacity participating in the adjustment of the auxiliary regulation system; embedding the standby capacity as a coupling variable into the three-stage rolling scheduling of day-ahead, day-ahead and real-time, and continuously transmitting the storage energy initial and final state of charge and the reservoir capacity boundary. The application makes the standby capacity dynamically match the prediction error and the hydropower adjustable capacity, reduces the storage energy demand capacity while maintaining the multi-time scale scheduling accuracy and the real-time power preservation rate.

Description

Technical Field

[0001] This invention relates to the field of new energy and power system dispatching technology, and in particular to multi-timescale joint dispatching technology for large-scale clean energy bases, specifically to a multi-timescale error compensation backup dispatching method for clean energy bases. Background Technology

[0002] In existing new power systems, large-scale clean energy bases, consisting of cascade hydropower stations, wind farms, photovoltaic power stations, energy storage devices, and transmission channels, have become an important form of absorbing new energy sources and ensuring power supply. However, because wind and photovoltaic power output is highly dependent on weather conditions, exhibiting intermittency, volatility, and uncertainty, there are deviations between actual and predicted outputs. These deviations vary significantly across different time scales, such as hours, quarter-hours, and minutes. To address these uncertainties, clean energy bases typically employ a framework of "multi-energy joint power forecasting + day-ahead, intraday, and real-time multi-timescale rolling dispatch," reserving a certain amount of standby capacity during dispatch to compensate for forecast errors.

[0003] For example, in the existing technology, such as the paper "ZHANG Q, SHUKLA A, XIE L. Efficient scenariogeneration for chance-constrained economic dispatch considering ambient wind conditions[J]. IEEE Transactions on Power Systems, 2024, 39(4): 5969-5980", a typical fixed reserve strategy is proposed. In the day-ahead and intraday dispatch phases, a fixed reserve coefficient is set based on experience, and the system reserve capacity is reserved according to a fixed ratio of system load to the predicted output of new energy sources (such as 10% of the predicted wind power value).

[0004] However, in the complex scenario of joint scheduling across multiple time scales in large-scale clean energy bases, the aforementioned fixed reserve strategy and existing scheduling technologies reveal the following significant technical shortcomings: First, existing technologies mostly reserve reserves based on fixed values ​​or proportions, which cannot dynamically match the significantly different prediction error levels of wind, solar, and load at different time scales. As the scheduling phase progresses from day-ahead to real-time, prediction accuracy improves and the error level is significantly reduced. If a fixed proportion is maintained, it will inevitably lead to an under-allocation of day-ahead resources (causing insufficient compensation, power shortages, and a decrease in power supply guarantee rate), while an over-allocation of real-time resources (crowding out the normal regulation space of hydropower and increasing the deep regulation pressure on energy storage and thermal power). Secondly, existing fixed reserve settings are often detached from reality, remaining only at the statistical demand level, lacking a constraint mapping with the actual physical adjustability of hydropower units. This leads to the dispatched reserve requirements easily exceeding the actual physical adjustability range of hydropower and becoming unfeasible, or causing the originally abundant rapid adjustment potential of hydropower to be idle. Third, if each stage is set up independently and the predicted value is substituted as a fixed value, it will lead to large vibrations in the unit output between stages, forming a chain-like deep regulation pressure from the day-ahead plan to the real-time operation. Fourth, the existing backup configuration method is difficult to accurately allocate backups by adjusting them upwards or downwards, thus failing to meet the stringent market technical requirements such as response time and duration to participate in system scheduling and allocation.

[0005] In summary, how to dynamically match the reserve capacity participating in the dispatch with the prediction error at each time scale and the physical adjustability of the hydropower unit, and continuously transfer it between day-ahead, intra-day, and real-time, thereby reducing the dependence on energy storage capacity while maintaining the dispatch accuracy at multiple time scales and the real-time power supply rate, is a technical problem that urgently needs to be solved in the field of clean energy base dispatch. Summary of the Invention

[0006] The existing fixed reserve strategy fails to dynamically match the reserve capacity with the physical adjustability of hydropower units as prediction errors vary across multiple time scales. This fixed reserve method, which lacks continuous transmission, results in large output vibrations, high pressure for deep regulation and ramp-up of units across different stages (especially thermal power). At the same time, this mode is highly dependent on energy storage capacity, making it difficult to reduce energy efficiency while simultaneously ensuring multi-time scale scheduling accuracy and real-time power supply rate.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A multi-timescale error compensation backup scheduling method for clean energy bases, including cascade hydropower stations, wind farms, photovoltaic power stations, energy storage devices, and transmission channels, is proposed. The method covers three timescales: day-ahead, intraday, and real-time, and sequentially executes the following steps: Step 1: Acquire historical data and predicted weather conditions, output power prediction values ​​at each time scale, and finally form the basic structure of the multi-time scale scheduling model; Step 2: Compare the power prediction value output in Step 1 with the historical operation, extract the absolute error distribution of prediction at each stage, and use this to solve the error compensation coefficient to prevent extreme predictions, and finally generate the error compensation demand boundary that reflects the uncertainty of the current time scale. Step 3: Using the error compensation demand boundary generated in Step 2, combined with the maximum ramping capacity and load limit at the hydropower physics level, perform multivariate lower bound cutoff operation to complete the setting of the single-station error compensation reserve capacity participating in auxiliary regulation, and divide the upper and lower regulation response capacity ranges. Step 4: Set the error compensation reserve capacity output in Step 3 as a cross-stage coupling variable, embed it into the rolling optimization of wider and narrower time scales in sequence to perform continuous boundary propagation, introduce network flow constraints for verification, and finally calculate and output the actual production scheduling instructions executed by the units in each stage.

[0008] Step 1 specifically includes the following steps: Sub-step 1-1) Obtain historical power data and meteorological data for wind power, solar power, hydropower, and load; generate and output predicted values ​​for wind power, solar power, hydropower, and load at various time scales: day-ahead, intraday, and real-time; record the time scale. Next period The predicted output and load demand for wind power and solar power are respectively: , , ; Sub-steps 1-2) Constructing the wind turbine output boundary: Establishing the nonlinear transformation relationship between the actual output power of the wind turbine and the wind speed state parameters to obtain the wind power physical output model; Sub-steps 1-3) Constructing the output boundary of the photovoltaic unit: In order to characterize the real power generation capacity of photovoltaic modules under complex weather conditions, an energy decay conversion mechanism between photovoltaic output and actual light intensity and operating temperature is established to obtain the photovoltaic output model; Sub-steps 1-4) Constructing the operation mechanism of cascade hydropower stations: Taking into account the upstream and downstream hydraulic connections and reservoir capacity constraints, determine the first... The output model and related constraints of the hydropower station are described by introducing the variables of discharge flow and head to characterize the actual physical capacity limitations of the hydropower unit; Sub-steps 1-5) combine the above-mentioned unit models to establish the overall power conservation operation framework of the system; introduce the energy storage charging and discharging state and perform power balance matching of the base, and finally output a complete multi-time scale scheduling model structure.

[0009] Step 2 specifically includes the following steps: Sub-step 2-1) Calculate the distribution of the absolute prediction error: For each time scale s and each type of energy k, subtract the predicted value output from sub-step 1-1 from the corresponding historical actual operating value sample by sample to obtain the absolute prediction error that characterizes the magnitude of uncertainty. Sub-step 2-2) Calculate the error compensation coefficient under the truncation probability constraint: In order to eliminate extreme small probability deviations, based on the distribution results of sub-step 2-1, extract the error threshold of the specified anti-extreme probability quantile and perform normalization mapping to generate the error compensation coefficients of each dimension. Sub-steps 2-3) Generate specific compensation demand constraints. By multiplying the error compensation coefficient by the current real-time forecast value, the abstract error probability is transformed into a hard demand indicator for the adjustment capacity, and finally the error compensation demand result for the corresponding period is output. By using the above methods to solve the error compensation coefficient and construct explicit error compensation demand, the reserve capacity demand can be dynamically matched with the magnitude of prediction error at different time scales such as day-ahead, intraday, and real-time. This can solve the technical defects of existing fixed reserves and the decoupling between time scale and prediction error, and improve the accuracy of multi-time scale scheduling.

[0010] Step 3 specifically includes the following steps: Sub-step 3-1) Perform physical cut-off operation on each cascade hydropower station at each time step: extract the error compensation requirement, the allowable adjustment extreme value of the maximum climbing rate of a single station, and the allowable adjustment extreme value based on the rated load limit. Take the definitive boundary of the data in these three dimensions, force the statistical requirement to be mapped back into the physical capacity envelope, and set the reserve capacity. Sub-step 3-2) Constructing the feasible domain of auxiliary regulation service response capacity: Based on the truncated reserve capacity obtained in sub-step 3-1, decouple and separate it according to the upward regulation and downward regulation attributes, and establish a capacity response boundary model in combination with the triggering conditions of auxiliary regulation response.

[0011] Step 4 specifically includes the following sub-steps: Sub-step 4-1) Perform continuous transfer of physical boundaries across scales: Set the final error compensation reserve capacity output from step 3 as the coupling mapping quantity of adjacent stages, and force the end load of the wide time scale planning and the reservoir capacity to be seamlessly used as the initial constraint starting point of the narrow time scale, and establish cascade boundaries. Sub-step 4-2) Introduce real-time network node security boundary verification: On the basis of the above continuous transmission, the inertial reserve of each unit and the transmission limit of the underlying power network of the system are further superimposed. When a local node does not meet the power support, the load is cut off and the security boundary is verified. Sub-step 4-3) Execute joint solution and output production scheduling instructions: Integrate the constraint variables of all stages, and use the maximization of power generation guarantee rate and the optimization of comprehensive power generation efficiency index as the driving mechanism to execute rolling solution calculation.

[0012] In sub-steps 1-2), the wind power physical output model is expressed as: (1); In the formula, Time scale Next period Wind power output; Wind speed; , , These are the cut-in wind speed, rated wind speed, and cut-out wind speed, respectively. This refers to the rated power of the wind turbine generator set; i Indicates the generator set unit number. w Indicates the physical properties of wind power; In sub-steps 1-3), the photovoltaic power output model is expressed as: (2); In the formula, Contribute to photovoltaic power; This represents the actual light intensity. Light intensity under standard test conditions; The power temperature coefficient; The operating temperature of photovoltaic modules; Temperature under standard test conditions; The rated power of the photovoltaic unit; PV represents the photovoltaic physical properties; r represents the rated physical parameters of the unit; stc represents the standard test condition baseline; and cell represents the physical node of the photovoltaic cell module. In sub-steps 1-4), the first The power output model of the hydroelectric power station is as follows: (3); In the formula, For the first Hydropower station on a time scale Next period contribution; This is the overall output coefficient; For power generation, the flow rate is released downstream; For water head; A collection of cascade hydropower stations; The relevant constraints are as follows: (4); In the formula, This represents the total outflow rate; This refers to the reservoir's water capacity. This refers to the inflow to the reservoir, and the discharge from the upstream power station, after a time lag, constitutes part of the inflow to the downstream power station; (with superscript) , These are the lower and upper limits of the corresponding variables; Time scale s The next time step; In sub-steps 1-5), the complete multi-timescale scheduling model is represented as follows: (5); In the formula, For energy storage charge; , These are the charging and discharging efficiencies, respectively. , These are the charging and discharging power, respectively. Powering wind power; Power output for thermal power plants; For load; For external power transmission; c and d represent the charging and discharging operation states of energy storage, respectively; L represents the system load attribute; and ex represents the physical attribute of cross-regional power transmission. Time scale s Next period t Total photovoltaic output.

[0013] In sub-step 2-1), the absolute prediction error is expressed as: (6); In the formula, For energy In time scale Next period , No. The absolute error of prediction for each sample; , These are the corresponding predicted output (demand) and actual output (demand); , This represents the total number of historical samples; data collection across various time scales and energy sources. The empirical distribution of its prediction absolute error is obtained, where f represents the individual index in the historical sample sequence, k represents the energy type classification identifier involved in regulation, and act represents the actual operational value collected by the system. In sub-step 2-2), the error compensation coefficient is expressed as: (7); In the formula, For energy In time scale The error compensation coefficient is a scalar corresponding to each time scale and each type of energy. The quantile function of the empirical distribution of absolute error; To prevent extreme prediction probability lower bounds, this embodiment takes... quantile; For energy The characteristic power. As can be seen from equation (7), the error compensation coefficient increases with the increase of the time scale. The wider the time scale (such as the day before), the more uncertain the prediction and the larger the error compensation coefficient; base represents the feature reference value used for normalization mapping; In sub-steps 2-3), the error compensation requirement is expressed as follows: (8); In the formula, Error compensation requirement is characterized as the regulation capacity required to compensate for the prediction errors of wind power, photovoltaics, and load at this time scale and during this period. This represents the predicted value for the corresponding energy.

[0014] In substep 3-1), the joint lower bound is represented as: (9); In the formula, For the first Hydropower station on a time scale Next period Error compensation reserve capacity participating in the regulation of the auxiliary control system; The error compensation requirement is allocated among the various hydropower stations by a distribution factor that satisfies... ; The error compensation requirements undertaken by the hydropower station; This represents the maximum rate of ascent for the hydropower station. The adjustable capacity corresponding to its maximum ramp rate within a certain period; This is the upper limit of the load (maximum output limit) of the hydropower station. The remaining adjustable capacity corresponding to its load limit. Equation (9) reduces the statistical error compensation requirement to the physically achievable adjustable range of the hydropower unit by taking the lower bound of the three, so that the obtained error compensation reserve capacity satisfies the error compensation requirement without exceeding the hydropower unit's ramp-up capability and output margin. By using the physical truncation method described above to remove the definitive boundary, the error compensation requirement in a purely statistical sense is precisely limited to the adjustable envelope of the hydropower unit that is physically realizable. This allows the reserve capacity to not only meet the deviation compensation requirements but also to have absolute feasibility. It transforms the rapid ramping capability of the hydropower unit into a reserve resource to accommodate uncertain prediction errors in a targeted and efficient manner. In sub-step 3-2), the capacity response boundary model is established as follows: (10); In the formula, , These represent adjustments to the error compensation reserve capacity, one upward and one downward. This is the baseline scheduling capacity obtained through system scheduling allocation; The performance contribution value of this reserve is included in the overall control effectiveness. Energy efficiency weighting coefficient; The scheduling response time is denoted by ; up and dn represent the physical directions of upward and downward adjustment of reserve capacity, respectively; cl represents the final physical boundary approved and confirmed by the system scheduling authority. s Determine the current time scale of the accounting; This completes the setting of the capacity boundary for participating in the regulation and reserve; By separating the reserve capacity into upper and lower limits and constructing response boundaries, the reserve capacity undertaken by the hydropower units can strictly meet the technical indicators of response time and duration stipulated in the auxiliary control service, and participate in the base control in a technically callable and verifiable manner, which greatly improves the utilization efficiency of reserve scheduling resources and the overall control efficiency of the system.

[0015] In sub-step 4-1), the cascade boundary is established using the following formula: (11); In the formula, and For adjacent, wider and narrower time scales; , These are the end of the previous stage and the beginning of the next stage, respectively. , The energy storage charge and storage capacity at the end of the previous stage serve as the initial boundary for the next stage. Narrow-scale time period The broad-scale time period to which it belongs; s' represents the adjacent broader and narrower time scales; end and start represent the end of the previous stage and the beginning of the next stage, respectively; In sub-step 4-2), the operation process is represented as follows: (12); In the formula, branch road The trend; branch road For nodes The sensitivity coefficient; The phase angle of the node; This is the upper limit of the tributary transmission capacity; For load shedding, When a node cannot obtain sufficient power under power flow constraints, it is included in the power outage area. "line" represents the physical attributes of the underlying power network transmission branch. l This indicates the network branch topology number, and the subscript j indicates the network node topology number; In sub-step 4-3), a rolling solution operation is performed, expressed by the following formula: (13); In the formula, This refers to the power generation guarantee rate (power supply guarantee rate). For base efficiency; , , , These are the priority dispatch weight parameters for wind power, photovoltaic power, hydropower, and thermal power, respectively. `max` represents the objective function driving the operation with the optimization direction of finding the maximum extremum within the physical boundary; to prevent the plans from becoming disconnected at different stages, output vibration and redundant lower limit constraints are further introduced for correction in the intraday rolling stage, expressed by the formula: (14); In the formula, This is the vibration amplitude coefficient of the unit output; This refers to within the day; Finally, the discharge flow of hydropower at all levels, the start-up and actual output of each generator unit are calculated to complete all scheduling and configuration processes; Through the synergistic effect of the above cross-scale physical boundary continuous transmission and joint solution mechanism, the reserve capacity can progressively reduce prediction errors at shorter time scales. This not only effectively smooths out the vibration amplitude of unit output between stages and greatly reduces the pressure of chain-like deep ramping on thermal power units, but also enables the entire system to maintain stable scheduling accuracy and good real-time power supply rate at multiple time scales without increasing or even reducing the dependence on energy storage physical capacity.

[0016] Compared with the prior art, the present invention has the following technical effects: 1) This invention solves the error compensation coefficient by inversely solving the probability lower bound of the empirical distribution of absolute error according to each time scale, and constructs an explicit error compensation requirement by multiplying the error compensation coefficient with the corresponding predicted value. This enables the reserve capacity to be matched with the magnitude of the prediction error at different time scales such as hours, quarter-hours, and minutes, overcoming the defect of existing fixed reserves being decoupled from time scales and prediction errors, and improving the accuracy of multi-time scale scheduling. 2) This invention sets the error compensation reserve capacity by taking the boundary of the error compensation requirement, the maximum climbing rate of the hydropower unit and the upper limit of the hydropower unit load. The statistical error compensation requirement is cut into the adjustable range that the hydropower unit can physically achieve, so that the obtained reserve capacity is both sufficient and feasible. The rapid climbing ability of the hydropower unit is transformed into a reserve resource to accommodate the prediction error. 3) This invention uses error compensation reserve capacity as a coupling variable to continuously transfer it across the stage boundaries of energy storage load and reservoir capacity in the day-ahead, intraday, and real-time stages. This allows the setting and transfer of reserve capacity to progressively reduce prediction errors on a shorter time scale, thereby reducing unit output vibration and chain-like deep regulation pressure between stages. The aforementioned error compensation coefficient setting, lower limit trimming, and cross-stage transfer are interconnected and work together to enable this invention to maintain multi-time scale scheduling accuracy and real-time power supply rate without increasing or even decreasing energy storage capacity. Its overall effect exceeds the sum of the individual effects of each link, thereby reducing dependence on energy storage capacity and reducing the deep regulation (climbing) pressure of thermal power units, making thermal power output more stable. 4) This invention separates the error compensation reserve capacity into upward and downward adjustments and participates in the clearing process in accordance with the response time and duration requirements of the auxiliary control system. This enables the error compensation reserve undertaken by hydropower to participate in the auxiliary service market in a technically applicable and callable form, and to be scheduled within the feasible domain according to the bid-winning capacity, thereby improving the utilization efficiency of reserve resources. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is an overall flowchart of the method of the present invention; Figure 2 This is a logic block diagram for error compensation reserve capacity setting in this invention; Figure 3 This is a schematic diagram of the clean energy base system structure of the present invention and the three-stage scheduling coupling relationship of day-ahead, intraday, and real-time scheduling. Figure 4 This is a timing diagram illustrating the three-stage rolling scheduling and stage boundary propagation of the present invention, which includes day-to-day and real-time phases. Figure 5 This is a comparison curve of the error compensation scheduling strategy and the fixed reserve strategy of the present invention in terms of output vibration error, power supply rate and energy storage capacity requirement. Figure 6 This is a comparison chart of the predicted curves and actual values ​​of wind power, photovoltaic power, load and power transmitted to other regions on a typical day under multiple time scales, including day-ahead, intraday and real-time, in an embodiment of the present invention. Figure 7 This is a bar chart comparing the error compensation coefficients of wind power, photovoltaic power, and load at three time scales: day-ahead, intraday, and real-time, in the embodiments of the present invention. Figure 8 The graphs are of the real-time stage error compensation reserve capacity and its error compensation requirement, the maximum hydropower ramping capacity, and the hydropower output margin in the embodiments of the present invention. Figure 9This is a real-time scheduling trajectory diagram of hydropower output, thermal power output, energy storage power, and energy storage load in an embodiment of the present invention. Detailed Implementation

[0018] like Figure 1 As shown, the overall process of the clean energy base multi-timescale error compensation reserve scheduling method proposed in this invention includes: multi-timescale power prediction and scheduling model construction, multi-timescale error compensation coefficient tuning, auxiliary control system error compensation reserve capacity tuning, and day-ahead-intraday-real-time three-stage rolling scheduling. In this embodiment, This indicates three time scales: day-to-day, intraday, and real-time, corresponding to hourly, quarter-hourly, and minute-level resolutions; Representing time scale The following period The duration of the time period at this scale; This refers to a cascade hydropower station. A collection of cascade hydropower stations; These represent wind power, photovoltaic power, and load, respectively.

[0019] Step 1: Construction of a multi-timescale power prediction and scheduling model; Step 1-1) Data Acquisition and Multi-Time Scale Prediction. Historical power data and meteorological data of wind power, photovoltaic (PV), hydropower, and load in the clean energy base are acquired. A joint prediction model for wind, PV, and load, incorporating Transformer encoding, feature attention, long short-term memory networks, and multi-task learning, is used to output predicted values ​​for wind power, PV, hydropower, and load at various time scales: day-ahead (hour), intraday (quarter-hour), and real-time (minute). Time scales are recorded. Next period The predicted output and load demand of wind power and solar power are respectively , , .

[0020] Steps 1-2) Wind power output model. The relationship between wind turbine output and wind speed is shown in equation (1): (1); In the formula, Time scale Next period Wind power output; Wind speed; , , These are the cut-in wind speed, rated wind speed, and cut-out wind speed, respectively. This refers to the rated power of the wind turbine.

[0021] Steps 1-3) Photovoltaic power output model. The photovoltaic unit output is shown in equation (2): (2); In the formula, Contribute to photovoltaic power; This represents the actual light intensity. Light intensity under standard test conditions; The power temperature coefficient; The operating temperature of photovoltaic modules; Temperature under standard test conditions; This refers to the rated power of the photovoltaic unit.

[0022] Steps 1-4) Cascade hydropower output model. The power output of the hydroelectric power station is shown in equation (3): (3); In the formula, For the first Hydropower station on a time scale Next period contribution; This is the overall output coefficient; For power generation, the flow rate is released downstream; The water head is the upper and lower limits of the reservoir capacity, upstream and downstream water levels, power generation discharge, total discharge, reservoir capacity, and water head. The mapping between water head and reservoir capacity, upstream and downstream water levels, as well as the upper and lower limits of the discharge flow, total discharge flow, reservoir capacity, and water head are shown in equation (4). (4); In the formula, This represents the total outflow rate; This refers to the reservoir's water capacity. This refers to the inflow to the reservoir, and the discharge from the upstream power station, after a time lag, constitutes part of the inflow to the downstream power station; (with superscript) , These are the lower and upper limits of the corresponding variables.

[0023] Steps 1-5) Energy storage model and power balance constraints. The energy storage device's charge and power balance are shown in equation (5): (5); In the formula, For energy storage charge; , These are the charging and discharging efficiencies, respectively. , These are the charging and discharging power, respectively. Powering wind power; Power output for thermal power plants; For load; This refers to the power transmitted externally.

[0024] Step 2: Tuning of multi-timescale error compensation coefficients; Step 2-1) Predict the absolute error distribution. Based on historical data, for each time scale... Each type of energy The predicted value is compared with the historical actual operating value to calculate the absolute error of the prediction. A historical sample on a time scale Next period The absolute error of the prediction is shown in equation (6): (6); In the formula, For energy In time scale Next period , No. The absolute error of prediction for each sample; , These are the corresponding predicted output (demand) and actual output (demand); , This represents the total number of historical samples. It includes data from various time scales and energy sources. The empirical distribution of its prediction absolute error is obtained.

[0025] Step 2-2) Tune the error compensation coefficients according to the lower probability limit. For each time scale... Each type of energy We take the probability lower bound of the empirical distribution of the absolute prediction error to prevent extreme predictions, and normalize it relative to the characteristic power of the energy source to obtain a scalar error compensation coefficient. As shown in equation (7): (7); In the formula, For energy In time scale The error compensation coefficient is a scalar corresponding to each time scale and each type of energy. The quantile function of the empirical distribution of absolute error; To prevent extreme prediction probability lower bounds, this embodiment takes... quantile; For energy The characteristic power. As can be seen from equation (7), the error compensation coefficient increases with the increase of the time scale. The wider the time scale (such as the day before), the more uncertain the prediction and the larger the error compensation coefficient.

[0026] Steps 2-3) Error Compensation Requirements. The time scale is obtained by multiplying the error compensation coefficient by the corresponding predicted value. Next period Error compensation requirements As shown in equation (8): (8); In the formula, Error compensation requirement is characterized as the regulation capacity required to compensate for the prediction errors of wind power, photovoltaics, and load at this time scale and during this period. This represents the predicted value for the corresponding energy.

[0027] Step 3: Setting the backup capacity for error compensation in the auxiliary control system; For each cascade hydropower station At each time step, the lower bound of the error compensation requirement, the adjustable capacity corresponding to the maximum ramp rate of the hydropower station, and the adjustable capacity corresponding to the upper limit of the load of the hydropower station are taken as the error compensation reserve capacity of the hydropower station participating in the auxiliary control system, as shown in equation (9): (9); In the formula, For the first Hydropower station on a time scale Next period Error compensation reserve capacity participating in the regulation of the auxiliary control system; The error compensation requirement is allocated among the various hydropower stations by a distribution factor that satisfies... ; The error compensation requirements undertaken by the hydropower station; This represents the maximum rate of ascent for the hydropower station. The adjustable capacity corresponding to its maximum ramp rate within a certain period; This is the upper limit of the load (maximum output limit) of the hydropower station. The remaining adjustable capacity corresponding to its load limit. Equation (9) cuts the statistical error compensation requirement to the adjustable range that the hydropower unit can physically realize by taking the lower bound of the three, so that the obtained error compensation reserve capacity can meet the error compensation requirement without exceeding the hydropower unit's climbing ability and output margin.

[0028] Furthermore, the error compensation reserve capacity is divided into upward and downward reserve, corresponding to the reserve capacity required for upward and downward compensation of prediction errors, respectively, and meeting the response time and duration requirements stipulated by the auxiliary control system. When organized in the form of an auxiliary service market, the declared error compensation reserve capacity participates in the system scheduling and allocation of the auxiliary control system, and the winning bid capacity obtained from the clearing process constrains the error compensation reserve capacity and the feasible domain of scheduling, as shown in equation (10): (10); In the formula, , These represent adjustments to the error compensation reserve capacity, one upward and one downward. This is the baseline scheduling capacity obtained through system scheduling allocation; The performance contribution value of this reserve is included in the overall control effectiveness. Energy efficiency weighting coefficient; The scheduling response time is denoted by ; up and dn represent the physical directions of upward and downward adjustment of reserve capacity, respectively; cl represents the final physical boundary approved and confirmed by the system scheduling authority. s Determine the current time scale of the accounting.

[0029] Step 4: Three-stage rolling scheduling and cross-stage coupling: day-to-day, intraday, and real-time scheduling; Step 4-1) Cross-stage coupling transfer. Incorporate error-compensated reserve capacity. As coupling variables, they are sequentially embedded into the rolling scheduling of the day-ahead, intraday, and real-time stages. The start-up plan, discharge flow, error compensation reserve capacity, and operating boundary determined in the previous stage (wider time scale) serve as constraints for the next stage (narrower time scale), and are continuously passed along with the stage boundaries of the energy storage start and end load and the reservoir capacity, as shown in Equation (11): (11); In the formula, and For adjacent, wider and narrower time scales; , These are the end of the previous stage and the beginning of the next stage, respectively. , The energy storage charge and storage capacity at the end of the previous stage serve as the initial boundary for the next stage. Narrow-scale time period The broad time period to which it belongs; specifically... , Each has a wider time scale s corresponding end of the phase Energy storage capacity and storage capacity; , They are respectively narrower time scales Corresponding to the initial period of the next stage Energy storage capacity and storage capacity; Step 4-2) Real-time inertia reserve and network power flow constraints. In the real-time phase, rotor inertia reserves for hydropower units and thermal power units, as well as virtual inertia reserves for wind power and photovoltaic units, are established, and network power flow constraints are introduced. The network power flow and outage area determination are shown in equation (12): (12); In the formula, branch road The trend; branch road For nodes The sensitivity coefficient; The phase angle of the node; This is the upper limit of the tributary transmission capacity; For load shedding, When a node cannot obtain sufficient power under power flow constraints, it is included in the power outage area.

[0030] Step 4-3) Optimization Objectives. Each stage is optimized on a rolling basis with the objectives of maximizing the power generation guarantee rate (power supply guarantee rate) and maximizing the base efficiency, as shown in equation (13): (13); In the formula, This refers to the power generation guarantee rate (power supply guarantee rate). For base efficiency; , , , These are the priority dispatch weight parameters for wind power, photovoltaic power, hydropower, and thermal power, respectively. The auxiliary control system efficiency is calculated according to formula (10) for error compensation backup.

[0031] During the intraday phase, based on equations (1) to (13), constraints on the unit output vibration range, ramp rate, and reserve redundancy are further introduced, as shown in equation (14): (14); In the formula, The vibration amplitude coefficient of the unit output; Equation (14) on the right indicates that the daily stage error compensation reserve capacity is not less than the apportionment value of the stage error compensation requirement, which constitutes the reserve redundancy constraint.

[0032] like Figure 2 As shown, the error compensation reserve capacity is obtained by taking the definitive value from the error compensation requirement, the maximum ramp rate of the hydropower unit, and the load limit. Figure 3 As shown, the clean energy base consists of cascade hydropower stations, wind farms, photovoltaic power stations, energy storage devices, transmission channels, and thermal power units, and operates through a three-stage scheduling system: day-ahead, intraday, and real-time.

[0033] like Figure 4As shown, in the day-ahead phase (e.g., starting at 0:00 every day), the start-up plan and discharge flow plan are solved on an hourly scale and the error compensation reserve capacity is output; in the intraday phase (e.g., starting every 4 hours), the day-ahead plan is inherited on a quarter-hour scale and the reserve capacity and grid connection scale are rolled over; in the real-time phase (e.g., starting every 0.5 hours), the real-time error distribution, energy storage call, inertia reserve and network flow constraints are taken into account on a minute scale, the unit output is refined and the power supply rate and power outage area are checked according to formula (12); in adjacent phases, the energy storage start and end charge and storage capacity boundary are continuously transferred through formula (11).

[0034] Example: This embodiment takes a four-stage cascade clean energy base in a river basin as the object. The base consists of four-stage cascade hydropower stations and their supporting wind farms, photovoltaic power stations, and energy storage devices. It is connected to the power system containing thermal power units through an external transmission channel for verification. The basic parameters such as system installed capacity, energy storage capacity, and external transmission channel capacity are shown in Table 1. The cost parameters such as electricity prices of wind power, photovoltaic, hydropower, thermal power, and energy storage / auxiliary control system are shown in Table 2. Among them, the total installed capacity of wind power and photovoltaic is about 1.83 times that of hydropower, falling within the range that wind and solar power installed capacity is 1.5 to 2.5 times that of hydropower installed capacity. The day-ahead, intraday, and real-time stages adopt hourly, quarter-hour, and minute-level resolutions respectively (the real-time step size in this embodiment is 5 minutes), and a typical day is selected for scheduling simulation.

[0035] Table 1. Basic parameters of the clean energy base system;

[0036] Table 2 Cost and Price Parameters;

[0037] Following step 1, multi-timescale power forecasting is performed for wind power, solar power, hydropower, and load, and a scheduling model is constructed. A comparison of the predicted curves and actual values ​​for wind power, solar power, load, and power transmission on a typical day at three timescales—day-ahead, intraday, and real-time—is shown below. Figure 6 As shown, the narrower the time scale, the higher the prediction resolution and the closer the predicted value is to the actual operating value.

[0038] Following step 2, the empirical distribution of the absolute prediction error at each time scale is statistically analyzed, and its 90th percentile is taken. The error compensation coefficients for each time scale and type of energy are obtained by inverse solution of equation (7). The results are shown in Table 3. Figure 7 As shown in Table 3 and Figure 7It can be seen that the error compensation coefficient of the same type of energy is greater than that of the same type of energy at a wider time scale (for example, the error compensation coefficient of wind power is 0.353, 0.203 and 0.163 at the day-ahead, intraday and real-time times respectively), which is consistent with the rule that the wider the time scale, the more uncertain the prediction. The error compensation requirement for each time scale is constituted by the product of the error compensation coefficient and the corresponding prediction value according to Equation (8).

[0039] Table 3 Error compensation coefficients for different time scales and energy types (taken as the 90th percentile of the empirical distribution of the absolute prediction error).

[0040] Following step 3, using equation (9), the lower bounds of the error compensation demand term, maximum ramp rate term, and load ceiling term for each cascade hydropower station are taken at each time step to obtain the error compensation reserve capacity participating in the auxiliary control system (auxiliary market) regulation. Taking the real-time stage as an example, the error compensation demand term, the maximum ramp rate term, the hydropower output margin term, and the error compensation reserve capacity obtained by taking the lower bounds of the three are as follows: Figure 8 As shown: When the error compensation requirement is large and the hydropower output margin or ramp-up capability is limited, the reserve capacity is correspondingly reduced to the range that the hydropower is physically achievable. This allows the reserve capacity to dynamically match the prediction error at each time scale with the physical adjustability of the hydropower, rather than remaining constant as with a fixed reserve baseline.

[0041] Step 4 involves using the error compensation reserve capacity as a coupling variable, continuously transferring and solving it across the day-ahead, intraday, and real-time stages via the initial and final energy storage load and reservoir capacity boundary. The actual scheduling trajectories of hydropower output, thermal power output, energy storage power, and energy storage load in the real-time stage are shown below. Figure 9 As shown: This invention replaces part of the energy storage regulation with error compensation backup for hydropower, resulting in more stable output of hydropower and thermal power, smaller fluctuations in energy storage load and no depletion, while under the fixed backup strategy, the energy storage load is quickly emptied during the evening load peak.

[0042] like Figure 5 As shown, the error compensation scheduling strategy of the present invention is compared with the scheduling strategy using fixed reserves under the same shared energy storage scale (1650MWh / 650MW). The main indicators are shown in Table 4, and the average absolute error of cross-stage unit output vibration is shown in Table 5.

[0043] Table 4 Comparison of key indicators between the present invention and the fixed backup strategy;

[0044] Table 5. Average absolute error of unit output vibration across stages;

[0045] From Table 4, Table 5 and Figure 5 It can be seen that: Firstly, regarding cross-stage output vibration, the combined average output vibration of hydropower and thermal power in this invention decreased from 56.7MW under the fixed reserve strategy to 45.3MW. Among them, the cross-stage output vibration of thermal power was significantly reduced (from 80.7MW to 40.1MW in the day-to-day range), and the average absolute error of the gradual ramp-up of thermal power output decreased from 33.9MW / step to 18.2MW / step, indicating that this invention effectively reduced the inter-stage unit output vibration and the pressure of deep regulation of thermal power. Among them, the hydropower output of this invention (35.6MW) in the day-to-real-time range is slightly higher than that of the fixed reserve strategy (29.4MW). This is because this invention actively takes over the prediction error by hydropower. The overall vibration and the deep regulation of thermal power are significantly better than the fixed reserve strategy. Secondly, regarding the real-time power supply rate, this invention achieves a real-time power supply rate of 100% under the shared energy storage scale, while the fixed reserve strategy is 99.61%, corresponding to an additional load shedding power of approximately 374.3MWh. Thirdly, regarding energy storage capacity requirements, since the error compensation reserve from hydropower replaces part of the energy storage regulation, the required energy storage capacity of this invention is 999.6 MWh and the required energy storage power is 457.5 MW, which are reduced by approximately 19.8% and 20.2% respectively compared to the fixed reserve strategy of 1245.8 MWh and 573.4 MW, thereby reducing the energy consumption for energy storage construction. The above results indicate that the error compensation coefficient tuning, lower bound trimming, and cross-stage transmission are interconnected and synergistic, enabling this invention to maintain multi-timescale scheduling accuracy and real-time power supply rate while reducing energy storage capacity requirements. Its overall effect exceeds the sum of the individual effects of each component.

[0046] In summary, this invention reduces the reliance on energy storage capacity while maintaining multi-timescale scheduling accuracy and real-time power supply rate. It achieves this by setting the error compensation coefficient based on the probability lower limit of the empirical distribution of absolute error predicted at each time scale, setting the error compensation reserve capacity in the auxiliary control system based on the error compensation demand, the hydropower unit's ramping capacity, and the load upper limit, and then continuously transferring this reserve capacity as a cross-stage coupling variable between the day-ahead, intra-day, and real-time stages. This invention has strong engineering application prospects.

Claims

1. A method for multi-timescale error compensation and backup scheduling of clean energy bases, wherein the clean energy bases include cascade hydropower stations, wind farms, photovoltaic power stations, energy storage devices, and transmission channels, characterized in that, This includes three time scales: previous day, intraday, and real-time, and the following steps are performed sequentially: Step 1: Acquire historical data and predicted weather conditions, output power prediction values ​​at each time scale, and finally form the basic structure of the multi-time scale scheduling model; Step 2: Compare the power prediction value output in Step 1 with the historical operation, extract the absolute error distribution of prediction at each stage, and use this to solve the error compensation coefficient to prevent extreme predictions, and finally generate the error compensation demand boundary that reflects the uncertainty of the current time scale. Step 3: Using the error compensation demand boundary generated in Step 2, combined with the maximum ramping capacity and load limit at the hydropower physics level, perform multivariate lower bound cutoff operation to complete the setting of the single-station error compensation reserve capacity participating in auxiliary regulation, and divide the upper and lower regulation response capacity ranges. Step 4: Set the error compensation reserve capacity output in Step 3 as a cross-stage coupling variable, embed it into the rolling optimization of wider and narrower time scales in sequence to perform continuous boundary propagation, introduce network flow constraints for verification, and finally calculate and output the actual production scheduling instructions executed by the units in each stage.

2. The method according to claim 1, characterized in that, Step 1 specifically includes the following steps: Sub-step 1-1) Obtain historical power data and meteorological data for wind power, solar power, hydropower, and load; generate and output predicted values ​​for wind power, solar power, hydropower, and load at various time scales: day-ahead, intraday, and real-time; record the time scale. Next period The predicted output and load demand for wind power and solar power are respectively: , , ; Sub-steps 1-2) Constructing the wind turbine output boundary: Establishing the nonlinear transformation relationship between the actual output power of the wind turbine and the wind speed state parameters to obtain the wind power physical output model; Sub-steps 1-3) Constructing the output boundary of the photovoltaic unit: In order to characterize the real power generation capacity of photovoltaic modules under complex weather conditions, an energy decay conversion mechanism between photovoltaic output and actual light intensity and operating temperature is established to obtain the photovoltaic output model; Sub-steps 1-4) Constructing the operation mechanism of cascade hydropower stations: Taking into account the upstream and downstream hydraulic connections and reservoir capacity constraints, determine the first... The output model and related constraints of the hydropower station are described by introducing the variables of discharge flow and head to characterize the actual physical capacity limitations of the hydropower unit; Sub-steps 1-5) combine the above-mentioned unit models to establish the overall power conservation operation framework of the system; introduce the energy storage charging and discharging state and perform power balance matching of the base, and finally output a complete multi-time scale scheduling model structure.

3. The method according to claim 1, characterized in that, Step 2 specifically includes the following steps: Sub-step 2-1) Calculate the distribution of the absolute prediction error: For each time scale s and each type of energy k, subtract the predicted value output from sub-step 1-1 from the corresponding historical actual operating value sample by sample to obtain the absolute prediction error that characterizes the magnitude of uncertainty. Sub-step 2-2) Calculate the error compensation coefficient under the truncation probability constraint: In order to eliminate extreme small probability deviations, based on the distribution results of sub-step 2-1, extract the error threshold of the specified anti-extreme probability quantile and perform normalization mapping to generate the error compensation coefficients of each dimension. Sub-steps 2-3) Generate specific compensation demand constraints. By multiplying the error compensation coefficient by the current real-time forecast value, the abstract error probability is transformed into a hard demand indicator for the adjustment capacity, and finally the error compensation demand result for the corresponding period is output. By using the above methods to solve the error compensation coefficient and construct explicit error compensation demand, the reserve capacity demand can be dynamically matched with the magnitude of prediction error at different time scales such as day-ahead, intraday, and real-time. This can solve the technical defects of existing fixed reserves and the decoupling between time scale and prediction error, and improve the accuracy of multi-time scale scheduling.

4. The method according to any one of claims 1 to 3, characterized in that, Step 3 specifically includes the following steps: Sub-step 3-1) Perform physical cut-off operation on each cascade hydropower station at each time step: extract the error compensation requirement, the allowable adjustment extreme value of the maximum climbing rate of a single station, and the allowable adjustment extreme value based on the rated load limit. Take the definitive boundary of the data in these three dimensions, force the statistical requirement to be mapped back into the physical capacity envelope, and set the reserve capacity. Sub-step 3-2) Constructing the feasible domain of auxiliary regulation service response capacity: Based on the truncated reserve capacity obtained in sub-step 3-1, decouple and separate it according to the upward regulation and downward regulation attributes, and establish a capacity response boundary model in combination with the triggering conditions of auxiliary regulation response.

5. The method according to claim 4, characterized in that, Step 4 specifically includes the following sub-steps: Sub-step 4-1) Perform continuous transfer of physical boundaries across scales: Set the final error compensation reserve capacity output from step 3 as the coupling mapping quantity of adjacent stages, and force the end load of the wide time scale planning and the reservoir capacity to be seamlessly used as the initial constraint starting point of the narrow time scale, and establish cascade boundaries. Sub-step 4-2) Introduce real-time network node security boundary verification: On the basis of the above continuous transmission, the inertial reserve of each unit and the transmission limit of the underlying power network of the system are further superimposed. When a local node does not meet the power support, the load is cut off and the security boundary is verified. Sub-step 4-3) Execute joint solution and output production scheduling instructions: Integrate the constraint variables of all stages, and use the maximization of power generation guarantee rate and the optimization of comprehensive power generation efficiency index as the driving mechanism to execute rolling solution calculation.

6. The method according to claim 2, characterized in that, In sub-steps 1-2), the wind power physical output model is expressed as: （1）； In the formula, Time scale Next period Wind power output; Wind speed; , , These are the cut-in wind speed, rated wind speed, and cut-out wind speed, respectively. This refers to the rated power of the wind turbine generator set; i Indicates the generator set unit number. w Indicates the physical properties of wind power; In sub-steps 1-3), the photovoltaic power output model is expressed as: （2）； In the formula, Contribute to photovoltaic power; This represents the actual light intensity. Light intensity under standard test conditions; The power temperature coefficient; The operating temperature of photovoltaic modules; Temperature under standard test conditions; The rated power of the photovoltaic unit; PV represents the photovoltaic physical properties; r represents the rated physical parameters of the unit; stc represents the standard test condition baseline; and cell represents the physical node of the photovoltaic cell module. In sub-steps 1-4), the first The power output model of the hydroelectric power station is as follows: （3）； In the formula, For the first Hydropower station on a time scale Next period contribution; This is the overall output coefficient; For power generation, the flow rate is released downstream; For water head; A collection of cascade hydropower stations; The relevant constraints are as follows: （4）； In the formula, This represents the total outflow rate; This refers to the reservoir's water capacity. This refers to the inflow to the reservoir, and the discharge from the upstream power station, after a time lag, constitutes part of the inflow to the downstream power station; (with superscript) , These are the lower and upper limits of the corresponding variables; Time scale s The next time step; In sub-steps 1-5), the complete multi-timescale scheduling model is represented as follows: （5）； In the formula, For energy storage charge; , These are the charging and discharging efficiencies, respectively. , These are the charging and discharging power, respectively. Powering wind power; Power output for thermal power plants; For load; For external power transmission; c and d represent the charging and discharging operation states of energy storage, respectively; L represents the system load attribute; and ex represents the physical attribute of cross-regional power transmission. Time scale s Next period t Total photovoltaic output.

7. The method according to claim 3, characterized in that, In sub-step 2-1), the absolute prediction error is expressed as: （6）； In the formula, For energy In time scale Next period , No. The absolute error of prediction for each sample; , These are the corresponding predicted output and actual output, respectively; , This represents the total number of historical samples; data collection across various time scales and energy sources. The empirical distribution of its prediction absolute error is obtained, where f represents the individual index in the historical sample sequence, k represents the energy type classification identifier involved in regulation, and act represents the actual operational value collected by the system. In sub-step 2-2), the error compensation coefficient is expressed as: (7); In the formula, For energy In time scale The error compensation coefficient is a scalar corresponding to each time scale and each type of energy. The quantile function of the empirical distribution of absolute error; To prevent extreme prediction probability lower bounds, this embodiment takes... quantile; For energy The characteristic power. As can be seen from equation (7), the error compensation coefficient increases with the increase of the time scale. The wider the time scale, the more uncertain the prediction and the larger the error compensation coefficient; base represents the feature reference value used for normalization mapping; In sub-steps 2-3), the error compensation requirement is expressed as follows: （8）； In the formula, Error compensation requirement is characterized as the regulation capacity required to compensate for the prediction errors of wind power, photovoltaics, and load at this time scale and during this period. This represents the predicted value for the corresponding energy.

8. The method according to claim 4, characterized in that, In substep 3-1), the joint lower bound is represented as: （9）； In the formula, For the first Hydropower station on a time scale Next period Error compensation reserve capacity participating in the regulation of the auxiliary control system; The error compensation requirement is allocated among the various hydropower stations by a distribution factor that satisfies... ; The error compensation requirements undertaken by the hydropower station; This represents the maximum rate of ascent for the hydropower station. The adjustable capacity corresponding to its maximum ramp rate within a certain period; This is the upper limit of the hydropower station's load. The remaining adjustable capacity corresponding to its load limit. Equation (9) reduces the statistical error compensation requirement to the physically achievable adjustable range of the hydropower unit by taking the lower bound of the three, so that the obtained error compensation reserve capacity satisfies the error compensation requirement without exceeding the hydropower unit's ramp-up capability and output margin. By using the physical truncation method described above to remove the definitive boundary, the error compensation requirement in a purely statistical sense is precisely limited to the adjustable envelope of the hydropower unit that is physically realizable. This allows the reserve capacity to not only meet the deviation compensation requirements but also to have absolute feasibility. It transforms the rapid ramping capability of the hydropower unit into a reserve resource to accommodate uncertain prediction errors in a targeted and efficient manner. In sub-step 3-2), the capacity response boundary model is established as follows: （10）； In the formula, , These represent adjustments to the error compensation reserve capacity, one upward and one downward. This is the baseline scheduling capacity obtained through system scheduling allocation; The performance contribution value of this reserve is included in the overall control effectiveness. Energy efficiency weighting coefficient; The scheduling response time is denoted by ; up and dn represent the physical directions of upward and downward adjustment of reserve capacity, respectively; cl represents the final physical boundary approved and confirmed by the system scheduling authority. s Determine the current time scale of the accounting; This completes the setting of the capacity boundary for participating in the regulation and reserve; By separating the reserve capacity into upper and lower limits and constructing response boundaries, the reserve capacity undertaken by the hydropower units can strictly meet the technical indicators of response time and duration stipulated in the auxiliary control service, and participate in the base control in a technically callable and verifiable manner, thereby greatly improving the utilization efficiency of reserve scheduling resources and the overall control efficiency of the system.

9. The method according to claim 1, 2, 4, 5, 6, 7, or 8, characterized in that, In sub-step 4-1), the cascade boundary is established using the following formula: （11）； In the formula, and For adjacent, wider and narrower time scales; , These are the end of the previous stage and the beginning of the next stage, respectively. , The energy storage charge and storage capacity at the end of the previous stage serve as the initial boundary for the next stage. Narrow-scale time period The broad time period to which it belongs; s' represents the adjacent wider and narrower time scales; end and start represent the end of the previous stage and the beginning of the next stage, respectively. In sub-step 4-2), the operation process is represented as follows: （12）； In the formula, branch road The trend; branch road For nodes The sensitivity coefficient; The phase angle of the node; This is the upper limit of the tributary transmission capacity; For load shedding, ; When a node cannot obtain sufficient power under power flow constraints, it is included in the power outage area; "line" represents the physical attributes of the underlying power network transmission branch. l This indicates the network branch topology number, and the subscript j indicates the network node topology number.

10. The method according to claim 9, characterized in that, In sub-step 4-3), a rolling solution operation is performed, expressed by the following formula: （13）； In the formula, This refers to the power generation guarantee rate (power supply guarantee rate). For base efficiency; , , , These are the priority dispatch weight parameters for wind power, photovoltaic power, hydropower, and thermal power, respectively. `max` represents the objective function driving the operation with the optimization direction of finding the maximum extremum within the physical boundary; to prevent the plans from becoming disconnected at different stages, output vibration and redundant lower limit constraints are further introduced for correction in the intraday rolling stage, expressed by the formula: （14）； In the formula, This is the vibration amplitude coefficient of the unit output; This refers to within the day; Finally, the discharge flow of hydropower at all levels, the start-up and actual output of each generator unit are calculated to complete all scheduling and configuration processes; Through the synergistic effect of the above cross-scale physical boundary continuous transmission and joint solution mechanism, the reserve capacity can progressively reduce prediction errors at shorter time scales. This not only effectively smooths out the vibration amplitude of unit output between stages and greatly reduces the pressure of chain-like deep ramping on thermal power units, but also enables the entire system to maintain stable scheduling accuracy and good real-time power supply rate at multiple time scales without increasing or even reducing the dependence on energy storage physical capacity.

Images