Energy storage system optimization scheduling method and device for high-frequency calling scenario

Abstract

The application discloses a high-frequency calling scene-oriented energy storage system optimization scheduling method and device, relates to the technical field of power system scheduling, and comprises the following steps: according to a power calling instruction sequence of an energy storage system, a high-frequency calling index is calculated; if the high-frequency calling index meets a predetermined high-frequency calling index, a sliding filter processing is performed on the power calling instruction sequence to obtain a high-frequency power component, a low-frequency energy component and a calling direction bias component; power type, energy type scheduling parameter configurations and SOC correction parameter configurations are respectively generated; and a high-frequency calling response strategy is obtained through global optimization. The application solves the technical problems that the existing energy storage scheduling mode cannot distinguish high-frequency fluctuations and continuous energy demand, cannot differentially allocate energy storage unit output, and simultaneously cannot balance state deviation caused by charging and discharging imbalance, and achieves the technical effect of realizing the optimization scheduling of response speed, equipment life and operation safety in the high-frequency power calling scene.

Description

Technical Field

[0001] This invention relates to the field of power system dispatching technology, and in particular to a method and apparatus for optimizing the dispatching of energy storage systems for high-frequency dispatch scenarios. Background Technology

[0002] Energy storage systems are widely used in high-frequency dispatch scenarios such as grid frequency regulation and new energy source fluctuation mitigation. Their operation and dispatch performance directly impact grid stability, energy storage device lifespan, and power supply reliability. Power command response control is a core component of energy storage operation, making optimized dispatch schemes crucial. Existing energy storage dispatch methods mostly employ a unified power allocation strategy to handle power commands, which can smoothly adapt to conventional low-frequency dispatch conditions. However, with the increasing demand for high-frequency and frequent reciprocating charging and discharging in new power systems, the shortcomings of traditional dispatch schemes are becoming increasingly apparent. Under high-frequency dispatch, power commands are a mixture of instantaneous fluctuations and continuous energy demands. Without component decomposition, it is impossible to differentiate and match energy storage units. Frequent charging and discharging easily causes unidirectional SOC offsets, and independent calculation of dispatch parameters for each unit can lead to conflicts. The resulting dispatch strategies suffer from poor comprehensiveness and insufficient accuracy, making it difficult to meet the management requirements for safe, efficient, and long-life operation of energy storage in high-frequency scenarios. Summary of the Invention

[0003] This application provides an optimized scheduling method and device for energy storage systems in high-frequency dispatch scenarios. It solves the technical problems of existing energy storage scheduling methods that cannot distinguish between high-frequency fluctuations and continuous energy demand, cannot differentiate the output of energy storage units, and are difficult to balance the state deviation caused by uneven charging and discharging, and have coupling conflicts of scheduling parameters in various dimensions.

[0004] The first aspect of this application provides an optimized scheduling method for energy storage systems in high-frequency dispatch scenarios. The method includes: calculating a high-frequency dispatch index based on a power dispatch command sequence of the energy storage system; if the high-frequency dispatch index meets a predetermined high-frequency dispatch index, performing sliding filtering on the power dispatch command sequence to obtain a high-frequency power component, a low-frequency energy component, and a dispatch direction bias component; acquiring the state data of each unit of the energy storage system, and performing power-type energy storage-driven scheduling on the energy storage system in conjunction with the high-frequency power component to obtain a power-type scheduling parameter configuration; performing energy-type energy storage-driven scheduling on the energy storage system based on the low-frequency energy component and the state data of each unit to obtain an energy-type scheduling parameter configuration; generating a State of Charge (SOC) correction parameter configuration based on the dispatch direction bias component, and performing global optimization on the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration to obtain a high-frequency dispatch response strategy.

[0005] A second aspect of this application provides an energy storage system optimization scheduling device for high-frequency dispatch scenarios. The device includes: a high-frequency dispatch index calculation module, used to calculate a high-frequency dispatch index based on a power dispatch command sequence of the energy storage system; a multi-type component acquisition module, which, if the high-frequency dispatch index meets a predetermined high-frequency dispatch index, performs sliding filtering on the power dispatch command sequence to obtain a high-frequency power component, a low-frequency energy component, and a dispatch direction bias component; a power-type parameter configuration acquisition module, used to acquire the status data of each unit of the energy storage system, and perform power-type energy storage drive scheduling on the energy storage system in conjunction with the high-frequency power component to obtain a power-type scheduling parameter configuration; an energy-type parameter configuration acquisition module, used to perform energy-type energy storage drive scheduling on the energy storage system based on the low-frequency energy component and the status data of each unit to obtain an energy-type scheduling parameter configuration; and a high-frequency dispatch response strategy acquisition module, used to generate a SOC correction parameter configuration based on the dispatch direction bias component, and perform global optimization on the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration to obtain a high-frequency dispatch response strategy.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application quantifies and identifies high-frequency operating conditions by analyzing energy storage power commands and decomposes them into three power components. It then matches energy storage units with appropriate characteristics to achieve branch output regulation, simultaneously corrects the energy storage state of charge based on charge and discharge offsets, and performs unified risk verification and global compensation on multiple sets of scheduling parameters. This adapts to the operational needs of energy storage systems under frequent reciprocating charge and discharge conditions, enabling energy storage scheduling strategies in high-frequency dispatch scenarios to balance response speed, equipment lifespan, and operational safety. This achieves the technical effect of optimized scheduling that balances response speed, equipment lifespan, and operational safety in high-frequency power dispatch scenarios. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a flowchart illustrating the energy storage system optimization scheduling method for high-frequency dispatch scenarios provided in this application embodiment.

[0009] Figure 2 This is a line graph showing the continuous trend of the state of charge (SBC) data of the energy storage units in the energy storage system optimized scheduling method for high-frequency dispatch scenarios provided in this application embodiment.

[0010] Figure 3 This is a schematic diagram of the structure of an energy storage system optimization scheduling device for high-frequency dispatch scenarios provided in this application embodiment.

[0011] Figure labeling: 1. High-frequency call index calculation module; 2. Multi-type component acquisition module; 3. Power-type parameter configuration acquisition module; 4. Energy-type parameter configuration acquisition module; 5. High-frequency call response strategy acquisition module. Detailed Implementation

[0012] This application provides an optimized scheduling method and device for energy storage systems in high-frequency dispatch scenarios. It solves the technical problems of existing energy storage scheduling methods that cannot distinguish between high-frequency fluctuations and continuous energy demand, cannot differentiate the output of energy storage units, and are difficult to balance the state deviation caused by uneven charging and discharging, and have coupling conflicts of scheduling parameters in various dimensions.

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0014] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0015] Example 1, as Figure 1 As shown, an optimized scheduling method for energy storage systems in high-frequency dispatch scenarios is described, wherein the method includes: Calculate high-frequency dispatch indicators based on the power dispatch command sequence of the energy storage system.

[0016] In this embodiment, the energy storage system is a complete set of power equipment integrating batteries, converters, and control units, capable of energy storage and bidirectional power output, used for grid frequency regulation and smoothing of new energy fluctuations. The power dispatch instruction sequence is a set of multiple control instructions containing the dispatch time and target output power, continuously issued and stored by the grid dispatching platform in chronological order.

[0017] Specifically, the calculation of high-frequency dispatch indicators is carried out based on the power dispatch command sequence of the energy storage system. First, the number of dispatches, average dispatch interval, and average power change rate corresponding to the power dispatch command sequence are counted. Next, the ratio of the number of dispatches to the predetermined number of dispatches is calculated as the first parameter of high-frequency dispatch, the ratio of the predetermined dispatch interval to the average dispatch interval is calculated as the second parameter of high-frequency dispatch, and the ratio of the average power change rate to the predetermined power change rate is calculated as the third parameter of high-frequency dispatch. Finally, the first, second, and third parameters of high-frequency dispatch are weighted according to the high-frequency dispatch weight conditions to obtain the high-frequency dispatch indicators. The specific implementation process of this step will be described in detail later.

[0018] If the high-frequency call index meets the predetermined high-frequency call index, the power call instruction sequence is subjected to sliding filtering to obtain the high-frequency power component, low-frequency energy component and call direction bias component.

[0019] In this embodiment, the high-frequency power component is the short-term, rapidly fluctuating power portion of the power dispatch command, used to match short-term, sudden increases or decreases in power demand, and is handled by power-type energy storage for rapid power throughput. The low-frequency energy component is the long-term, stable base power portion of the power dispatch command, corresponding to continuous and stable charging and discharging energy demand, and is carried out long-term by energy-type energy storage. The dispatch direction bias component is the difference between the accumulated energy of charging and discharging within a period, reflecting whether the overall dispatch is biased towards charging or discharging, and is used to correct the overall state of charge level of the energy storage.

[0020] Optionally, after calculating the high-frequency dispatch index, the real-time calculated high-frequency dispatch index is compared with a pre-stored predetermined high-frequency dispatch index. The predetermined high-frequency dispatch index is pre-set by technicians based on the high-frequency operating condition judgment standards for the energy storage system's grid connection. The high-frequency dispatch index is a normalized value, ranging from 0 to 1. An example of the predetermined high-frequency dispatch index can be configured as 0.6, and technicians can flexibly adjust it within the range of 0.5 to 0.8 according to the actual operating condition definition requirements. After setting, it is stored in the controller's local storage unit. When the high-frequency dispatch index value is greater than or equal to the predetermined high-frequency dispatch index, it is determined that the current dispatch condition meets the high-frequency dispatch characteristics, and the subsequent sliding filter processing procedure is initiated; when the high-frequency dispatch index value is less than the predetermined high-frequency dispatch index, the sliding filter splitting procedure is not initiated, and the preset basic dispatch method is executed.

[0021] When the high-frequency call index reaches the predetermined high-frequency call index, the length of the sliding filter window is determined based on the high-frequency call index. Based on this window length, all power call instructions within the current sliding filter window corresponding to the current scheduling time are extracted from the power call instruction sequence. First, the low-frequency energy component is obtained based on the target call power of each instruction within the window. Then, the high-frequency power component is obtained by subtracting the low-frequency energy component from the target call power at the current time. At the same time, the cumulative call energy of charging and discharging within the window is counted and the difference between the two is calculated to obtain the call direction bias component. The specific implementation process of this step will be described in detail later.

[0022] The status data of each unit of the energy storage system is obtained, and the energy storage system is power-driven energy storage scheduling is performed in combination with the high-frequency power component to obtain the power-driven scheduling parameter configuration.

[0023] In one embodiment of this application, the state data of each unit of the energy storage system is obtained and the high-frequency response capability vector of each energy storage unit is extracted. Based on this, a set of power-type energy storage candidate units is obtained. After obtaining the power-type energy storage priority conditions through scheduling priority evaluation, scheduling decisions are made on the candidate units in combination with high-frequency power components to obtain a power-type energy storage scheduling scheme. After amplitude limiting verification and lifetime loss compensation processing, the power-type scheduling parameter configuration is finally generated. The specific implementation process of this step will be described in detail later.

[0024] Based on the low-frequency energy components and the state data of each unit, the energy storage system is subjected to energy-type energy storage drive scheduling to obtain energy-type scheduling parameter configuration.

[0025] Specifically, based on the state data of each unit of the energy storage system, the low-frequency response capability vector corresponding to each energy storage unit is extracted. This is used to screen and obtain a set of energy-type energy storage candidate units. After determining the priority conditions of energy-type energy storage through scheduling priority evaluation, the scheduling decision is completed by combining the low-frequency energy components to obtain the energy-type energy storage scheduling scheme. Then, the risk prediction feedforward compensation operation is performed on the scheduling scheme to finally generate the corresponding energy-type scheduling parameter configuration. The specific implementation process of this step will be described in detail later.

[0026] Based on the call direction bias component, a SOC correction parameter configuration is generated, and the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration are globally optimized to obtain a high-frequency call response strategy.

[0027] In this embodiment, SOC, State of Charge, is the ratio of the remaining usable capacity of the energy storage battery to its rated total capacity, used to intuitively reflect the current remaining power level of the battery.

[0028] Specifically, the SOC feature matrix is ​​extracted based on the state data of each unit of the energy storage system. The influence of the SOC feature matrix is ​​mapped by combining the call direction bias component to obtain the SOC change path. An initial correction strategy is generated based on the SOC change path, and safety boundary optimization and high-frequency response concession compensation are carried out on the initial correction strategy to finally obtain the SOC correction parameter configuration. The specific implementation process of this step will be described in detail later.

[0029] Next, the candidate response scheduling strategy is obtained by integrating the power-type scheduling parameter configuration, energy-type scheduling parameter configuration, and SOC correction parameter configuration. The candidate response scheduling strategy is then used to predict scheduling conflict accidents and reserve margin risk accidents to obtain two types of accident chains. Based on the two types of accident chains, global compensation is then applied to the candidate response scheduling strategy to finally generate a high-frequency call response strategy. The specific implementation process of this step will also be described in detail later.

[0030] Furthermore, the method provided in this application embodiment includes: The system calculates the number of calls, average call interval, and average power change rate of the power call instruction sequence; outputs the ratio of the number of calls to the predetermined number of calls as the first parameter of high-frequency call; outputs the ratio of the predetermined call interval to the average call interval as the second parameter of high-frequency call; outputs the ratio of the average power change rate to the predetermined power change rate as the third parameter of high-frequency call; and performs a weighted calculation on the first parameter, the second parameter, and the third parameter of high-frequency call according to the high-frequency call weight condition to obtain the high-frequency call index.

[0031] Specifically, during operation, the energy storage system continuously receives real-time power control signals from the superior power grid dispatching platform. The system's built-in signal acquisition module continuously captures dispatch signals at millisecond sampling intervals. Each time a new power control signal is acquired, the current dispatch time and corresponding target output power value are recorded, and the instruction is appended to a fixed-duration rolling storage queue. This queue uses a single-step sliding update method; each time a new instruction is added, the oldest instruction data in the queue is automatically removed, maintaining a constant statistical duration window covered by the queue. This duration window can be pre-configured according to the power grid frequency regulation service specifications, for example, it can be configured to 5 minutes. All instructions in the queue are stored in order of their signal generation time, forming a continuous and complete power call instruction sequence within the corresponding window duration. Each time high-frequency call index calculation is initiated, the complete power call instruction sequence in the current rolling queue is read for subsequent statistical calculations.

[0032] Next, after obtaining the complete power call instruction sequence within the specified duration window, the system iterates through and reads all the stored instruction data within the power call instruction sequence, counting each instruction to obtain the total number of calls contained in the sequence; it then extracts the time difference between adjacent instructions, sums all time differences, and calculates the arithmetic mean to obtain the average call interval; it then calculates the power difference between adjacent instructions, takes the absolute value of the power difference, divides it by the time interval between the two instructions, and obtains the power change rate of a single interval; finally, it sums the power change rates of all intervals within the window and calculates the arithmetic mean to obtain the average power change rate. The above steps uniformly use absolute values ​​to calculate the power change rate, which avoids the numerical cancellation problem caused by the positive and negative values ​​of the charging and discharging directions, ensuring that the statistical results objectively reflect the severity of power fluctuations.

[0033] The predetermined number of calls, predetermined call interval, and predetermined power change rate are all preset fixed reference thresholds. The statistical duration of these three thresholds is completely consistent with the rolling storage queue duration window corresponding to the power call instruction sequence. Specifically, taking a regional power grid frequency regulation service standard as an example, when the statistical duration window is configured as 5 minutes, the predetermined number of calls can be set to 10 times, corresponding to the maximum frequency of regular dispatch under the same statistical duration; the predetermined call interval can be set to 30 seconds, corresponding to the minimum time interval for issuing regular low-frequency dispatch instructions under the same statistical duration; and the predetermined power change rate can be set to 0.2 pu / s, corresponding to the rated allowable stable power change rate of the energy storage device. These values ​​can be flexibly modified by those skilled in the art according to the power grid requirements of different projects. The setting process relies on manual input and saving through the controller's human-machine interface. Each time the high-frequency call index calculation step is executed, the controller automatically reads the three fixed thresholds from the local storage unit to participate in the calculation.

[0034] After completing the reading of various statistical values ​​and preset thresholds, the system performs three sets of ratio calculation operations: dividing the number of calls obtained by statistics by the predetermined number of calls to obtain the first parameter of high-frequency calls; dividing the predetermined call interval by the average call interval obtained by statistics to obtain the second parameter of high-frequency calls; and dividing the average power change rate obtained by statistics by the predetermined power change rate to obtain the third parameter of high-frequency calls.

[0035] Subsequently, the pre-configured high-frequency call weight conditions are retrieved. These conditions include fixed weight coefficients corresponding to the first, second, and third parameters of the high-frequency call, namely, the first, second, and third weights. The specific configuration process will be detailed in subsequent embodiments. All weight coefficients range from 0 to 1, and the sum of the three weight coefficients is always 1. For example, the first high-frequency call weight can be configured as 0.3, the second as 0.3, and the third as 0.4. It should be noted that the specific values ​​of these weight coefficients can be pre-set by technicians based on the actual high-frequency operating conditions of the energy storage system connected to the grid. For instance, if the grid frequency regulation service focuses more on the rate of power change, the value of the third high-frequency call weight can be appropriately increased. Each weight coefficient can have a boundary value of 0 or 1, but the sum of the three must be 1. Once set, the weight coefficients are permanently stored in the controller.

[0036] After determining the above parameter values, calculate the high-frequency call index using the following formula: Where H represents the high-frequency call metric, and N represents the number of calls. Indicates the predetermined number of calls. Indicates the scheduled call interval. Indicates the average call interval. Indicates the average power change rate. Let α represent the predetermined power change rate, β represent the first weight of high-frequency call, γ represent the second weight of high-frequency call, and α + β + γ = 1. The final summation value H is the high-frequency call index.

[0037] The system automatically initiates execution after the energy storage controller detects a rolling window update of the power dispatch command sequence. It aggregates dispatch commands within a fixed-duration window to form a complete sequence, statistically analyzes command operation characteristics using a unified standard, calculates characteristic parameters by matching them with preset reference thresholds based on the same time base, and then performs a weighted calculation with fixed weights to obtain a unified evaluation value. This allows for a quantitative determination of the high-frequency nature of the current dispatching operation.

[0038] Furthermore, the method provided in this application embodiment includes: The high-frequency call weighting conditions include a first high-frequency call weight, a second high-frequency call weight, and a third high-frequency call weight.

[0039] Optionally, the high-frequency call weighting condition includes a first high-frequency call weight, a second high-frequency call weight, and a third high-frequency call weight. The first high-frequency call weight corresponds to the degree of influence of the first high-frequency call parameter in the calculation of the high-frequency call metric; the second high-frequency call weight corresponds to the degree of influence of the second high-frequency call parameter in the calculation of the high-frequency call metric; and the third high-frequency call weight corresponds to the degree of influence of the third high-frequency call parameter in the calculation of the high-frequency call metric.

[0040] The specific values ​​for the first, second, and third weights of high-frequency call are pre-configured by technicians based on the actual frequency regulation requirements of the energy storage system connected to the grid. During configuration, technicians first rank the importance of three indicators—call frequency, average call interval, and average power change rate—in determining high-frequency call conditions, and then assign values ​​according to the ranking. The values ​​of the three weight coefficients range from 0 to 1, with each coefficient allowed to take boundary values ​​of 0 or 1. The sum of the three is always 1, ensuring a uniform quantitative scale for the weighted calculation and adapting to the subsequent high-frequency call condition determination threshold. After technicians complete the numerical input, the controller's built-in parameter verification program automatically performs a summation check on the three weight coefficients. If the summation result is not equal to 1, a parameter anomaly warning pops up, requiring technicians to adjust the values ​​until the normalization constraints are met.

[0041] Taking a regional power grid frequency regulation application scenario as an example, the power frequency regulation business in this region is characterized by frequent power fluctuations. Technicians determined that the average power change rate is the most critical indicator of high-frequency dispatch conditions, followed by the number of dispatches, and then the average dispatch interval. Based on this determination, the first weight for high-frequency dispatch is configured as 0.3, the second weight as 0.3, and the third weight as 0.4, with the sum of the three weight coefficients being 1, satisfying the normalization constraint. If another energy storage system is connected to the power grid and the frequency of grid dispatch commands is extremely high, technicians determined that the number of dispatches is the most critical indicator of high-frequency dispatch conditions. In this case, the first weight for high-frequency dispatch can be configured as 0.5, the second weight as 0.2, and the third weight as 0.3.

[0042] After the weighting coefficients are configured, maintenance personnel with system configuration permissions manually input and save them to the controller's local storage unit through the energy storage controller's human-machine interface. Each time a high-frequency call index calculation is performed, the controller automatically reads three sets of weighting coefficients from the local storage unit to participate in the weighted summation calculation. The weighting parameters can be modified according to the needs of power grid business adjustments; however, modifications require authorization verification before execution. Under normal operating conditions, the weighting parameters remain fixed to prevent accidental operations from affecting the stability of the index calculation.

[0043] Furthermore, the method provided in this application embodiment includes: The sliding filter window length is determined based on the high-frequency call index; based on the sliding filter window length, the current sliding filter window corresponding to the current scheduling time is determined in the power call instruction sequence, and multiple corresponding power call instructions within the current sliding filter window are extracted; sliding filter calculation is performed based on the target call power of each corresponding power call instruction to obtain the low-frequency energy component; the difference between the current target call power corresponding to the current scheduling time and the low-frequency energy component is calculated to obtain the high-frequency power component; based on the multiple corresponding power call instructions, the cumulative call energy in the discharge direction and the cumulative call energy in the charging direction are calculated; the difference between the cumulative call energy in the discharge direction and the cumulative call energy in the charging direction is calculated to obtain the call direction bias component.

[0044] Specifically, the length of the sliding filter window is dynamically determined based on the value of the high-frequency call index, and the length of the sliding filter window is negatively correlated with the high-frequency call index. In practice, the controller pre-stores multiple sets of mapping relationships between high-frequency call index value ranges and corresponding sliding filter window lengths. The higher the high-frequency call index value, the shorter the corresponding sliding filter window length; the lower the high-frequency call index value, the longer the corresponding sliding filter window length. The sliding filter window length is measured in units of the number of power call commands it contains. For example, when the high-frequency call index is in the range of 0.6 to 0.7, the sliding filter window length corresponds to 20 power call commands; when the high-frequency call index is in the range of 0.7 to 0.8, the sliding filter window length corresponds to 15 power call commands; and when the high-frequency call index is greater than 0.8, the sliding filter window length corresponds to 10 power call commands. The mapping relationship is pre-configured and stored by technicians based on the characteristics of power grid fluctuations. After the controller reads the current high-frequency call index and matches the corresponding range, it can determine the sliding filter window length used for this calculation.

[0045] After determining the length of the sliding filter window, the current scheduling time is used as the window's time endpoint. A corresponding number of power call instructions are traced back in the power call instruction sequence to define the current sliding filter window corresponding to the current scheduling time. The controller iterates through all instruction data within the power call instruction sequence, filters all power call instructions falling within the current sliding filter window range based on timestamp matching, and extracts the target power value and instruction timestamp information for each power call instruction to form the current window's computational dataset for subsequent component calculations.

[0046] The low-frequency energy component is calculated using a moving average filter. The controller reads the target call power values ​​of all power call commands within the current sliding filter window, performs a summation operation on all target call power values, and then divides the summation result by the total number of power call commands contained in the window to obtain the arithmetic mean of the power within the window. This average value is the low-frequency energy component corresponding to the current scheduling time.

[0047] After calculating the low-frequency energy component, the current target call power value corresponding to the current scheduling moment is extracted. The low-frequency energy component value is then subtracted from the current target call power value, and the difference is the high-frequency power component. The positive or negative value of the high-frequency power component represents the direction of deviation of the instantaneous fluctuation from the energy baseline, and the absolute value represents the severity of the instantaneous fluctuation.

[0048] Next, the controller iterates through all power call commands within the current sliding filter window, distinguishing between discharge direction commands and charging direction commands based on their positive or negative power attributes, and performs energy accumulation calculations for both types of commands. The duration time interval corresponding to a single command is determined as follows: for all power call commands within the current sliding filter window except the last one, the duration time interval is the difference between the timestamp of that command and the timestamp of the next adjacent command; for the last power call command within the window, the duration time interval is the difference between the current scheduling time and the timestamp of that command. The call energy of a single command is obtained by multiplying the absolute value of the target call power by the duration time interval corresponding to that command. The call energy of all discharge direction commands is accumulated to obtain the cumulative call energy in the discharge direction, and the call energy of all charging direction commands is accumulated to obtain the cumulative call energy in the charging direction. After completing the calculation of the two types of cumulative energy, the cumulative call energy in the discharge direction is subtracted from the cumulative call energy in the charging direction, and the difference obtained is the call direction bias component.

[0049] By setting a high-frequency operating condition triggering start-up mechanism, adaptively matching and adjusting the filter window length, and using the moving average method to separate high and low frequency power components, the directional bias is obtained by synchronously calculating the cumulative energy difference between charging and discharging within the window. This achieves the effect of accurately separating different characteristic components in the power command and providing accurate input basis for subsequent energy storage scheduling by type.

[0050] Furthermore, the method provided in this application embodiment includes: The length of the sliding filter window is negatively correlated with the high-frequency call index.

[0051] Specifically, setting the negative correlation between the sliding filter window length and the high-frequency dispatch index can adaptively adjust the granularity of the filtering process for different levels of high-frequency dispatch conditions, and optimize the splitting effect of high-frequency power components, low-frequency energy components and dispatch direction bias components.

[0052] When the high frequency call index value is high, it means that the fluctuation frequency and change amplitude of the current power call command are at a high level. At this time, matching a shorter sliding filter window length can reduce the smoothing effect of the moving average calculation, improve the response sensitivity of high frequency power component extraction, accurately capture the details of instantaneous power fluctuation, and avoid the high frequency features being masked due to excessive smoothing of the long window.

[0053] When the high-frequency call index value is low, it means that the current power call command is generally stable. At this time, matching a longer sliding filter window length can enhance the smoothing effect of the filter, stably extract the baseline of the low-frequency energy component, reduce the impact of random small disturbances on the component splitting results, and improve the stability of the low-frequency energy component calculation results.

[0054] The negative correlation matching mechanism enables the power component splitting process to dynamically adapt to the fluctuation characteristics of the current scheduling conditions, taking into account both the response speed and computational stability of component extraction, and providing reliable data support for the subsequent precise scheduling of energy storage units of different types.

[0055] Furthermore, the method provided in this application embodiment includes: Based on the status data of each unit, high-frequency response capability features are extracted from each energy storage unit in the energy storage system to obtain multiple high-frequency response capability vectors. Based on these vectors, power-type energy storage scheduling feasibility is screened for each energy storage unit to obtain a set of power-type energy storage candidate units. Scheduling priority is evaluated on the set of power-type energy storage candidate units to obtain power-type energy storage priority conditions. Based on these priority conditions, scheduling decisions are made on the set of power-type energy storage candidate units according to the high-frequency power components to obtain a power-type energy storage scheduling scheme. Finally, the power-type energy storage scheduling scheme is subjected to amplitude limiting verification and lifetime loss compensation to generate the power-type scheduling parameter configuration.

[0056] Specifically, the energy storage controller collects real-time status data of all energy storage units within the energy storage system. This data includes the current state of charge, maximum allowable charge / discharge power, power response delay, remaining cycle life, and current operating temperature for each unit. After data acquisition, characteristic parameters related to high-frequency response capability are extracted for each energy storage unit. Three key parameters—power response delay, current available charge / discharge power margin, and power regulation rate—are selected and their corresponding values ​​are directly read from the collected unit status data. These three values ​​are then combined in a fixed order to form a high-frequency response capability vector for each individual energy storage unit. Multiple high-frequency response capability vectors are generated for all energy storage units.

[0057] After constructing all high-frequency response capability vectors, a feasibility screening operation for power-type energy storage scheduling is performed. The controller pre-stores the qualification thresholds corresponding to three characteristic parameters: the upper limit threshold for power response delay time, the lower limit threshold for available charge / discharge power margin, and the lower limit threshold for power regulation rate. The controller iterates through each set of high-frequency response capability vectors, comparing the three characteristic values ​​within the vector with the corresponding qualification thresholds one by one. When all three characteristic values ​​of a certain energy storage unit meet the threshold requirements, the energy storage unit is included in the power-type energy storage candidate unit set; any energy storage unit that does not meet any threshold requirement is not included in the power-type energy storage candidate unit set.

[0058] For all energy storage units within the candidate set of power-type energy storage units, a scheduling priority evaluation calculation is performed. The controller pre-configures priority weight coefficients for three characteristic parameters. These weight coefficients are set based on the dependence of the high-frequency response on each characteristic, with power response delay having the highest weight, followed by power regulation rate, and available charge / discharge power margin having the lowest weight. For example, the weight for power response delay can be configured as 0.5, power regulation rate as 0.3, and available charge / discharge power margin as 0.2, with a sum of 1. During the calculation, an extreme value normalization method is used to normalize each characteristic parameter. For positive indicators such as available charge / discharge power margin and power regulation rate, the normalized value is the current characteristic value divided by the maximum value of that characteristic within the candidate set; for negative indicators such as power response delay, the normalized value is the minimum value of that characteristic within the candidate set divided by the current characteristic value. The normalized feature values ​​are multiplied by their corresponding weights, and the sum of the three products is used to obtain the priority comprehensive score of a single energy storage unit. After all candidate units have been calculated, they are sorted in descending order of comprehensive score. This sorting rule is the priority condition for power-type energy storage.

[0059] Next, after obtaining the power-type energy storage priority conditions, scheduling and allocation are performed based on the total power demand of the high-frequency power components. The controller reads the current high-frequency power component values ​​to determine the total charging and discharging power value and power direction to be allocated. Positive high-frequency power components correspond to the power demand in the discharging direction, while negative values ​​correspond to the power demand in the charging direction. Power is allocated sequentially according to the power-type energy storage priority conditions, starting with the highest priority energy storage unit. The highest priority unit is allocated its currently available maximum charging and discharging power first. If there is still a surplus in the total power demand, power is allocated to the next priority energy storage unit until the power demand of all high-frequency power components is completely covered. All participating energy storage units and their corresponding target power together constitute the power-type energy storage scheduling scheme. If the sum of the available charging and discharging power margins of all energy storage units in the current candidate unit set still cannot cover the total power demand of the high-frequency power components, then the maximum available power of each energy storage unit is allocated. The insufficient portion is marked as a power deficit and reported by the energy storage system to the grid dispatch platform for processing.

[0060] After obtaining the preliminary power-type energy storage scheduling scheme, limit verification and lifetime loss compensation are performed sequentially. In the limit verification phase, the allocated power of each energy storage unit is checked against the upper and lower limits of its current state of charge (SBC). If the allocated power exceeds the limits, the target power is adjusted to within the limits, and the resulting power difference is compensated by subsequent energy storage units according to priority. In the lifetime loss compensation phase, the cumulative cycle count of each energy storage unit is read. Energy storage units with a cumulative cycle count reaching 70% of their rated cycle life are considered to have a high cumulative cycle count. The change in the target power allocated to this energy storage unit in the current scheduling cycle compared to the actual output power in the previous scheduling cycle is limited to within 80% of the original change, i.e., the change in a single power adjustment is reduced by 20%, reducing the frequency of frequent charge-discharge switching and minimizing cycle loss accumulation. After completing these two corrections, the final target power, action execution sequence, and other parameters of each energy storage unit are compiled to generate the power-type scheduling parameter configuration.

[0061] By constructing capability vectors through feature extraction, determining candidate units through threshold screening, clarifying scheduling priorities through weighted scoring, and then allocating high-frequency power demand in sequence with corresponding limiting and loss compensation processing, the energy storage unit can be ensured to operate safely and delay equipment wear while responding quickly to high-frequency power fluctuations.

[0062] Furthermore, the method provided in this application embodiment includes: Based on the status data of each unit, low-frequency response capability features are extracted from each energy storage unit in the energy storage system to obtain multiple low-frequency response capability vectors. Based on these multiple low-frequency response capability vectors, energy storage scheduling feasibility is screened for each energy storage unit to obtain a set of energy storage candidate units. Scheduling priority is evaluated on the set of energy storage candidate units to obtain energy storage priority conditions. Based on these energy storage priority conditions, scheduling decisions are made on the set of energy storage candidate units according to the low-frequency energy components to obtain an energy storage scheduling scheme. Risk prediction feedforward compensation is applied to the energy storage scheduling scheme to generate the energy storage scheduling parameter configuration.

[0063] In one embodiment, the energy storage controller collects real-time status data of all energy storage units within the energy storage system. This data includes the current state of charge, rated energy storage capacity, charge-discharge conversion efficiency, rated continuous charge-discharge power, battery health status, and current operating temperature for each energy storage unit. After data acquisition, characteristic parameters related to low-frequency response capability are extracted for each energy storage unit. Three characteristic parameters are selected: current available energy storage capacity, charge-discharge conversion efficiency, and continuous charge-discharge support duration. The continuous charge-discharge support duration is calculated by dividing the current available energy storage capacity by the rated continuous discharge power of the energy storage unit. The rated discharge power serves as a benchmark to reflect the unit's ability to stably handle continuous energy dispatch for an extended period. These three characteristic values ​​are directly read from or calculated from the collected unit status data. These three values ​​are combined in a fixed order to form a low-frequency response capability vector corresponding to a single energy storage unit. Multiple sets of low-frequency response capability vectors are generated for all energy storage units.

[0064] After constructing all low-frequency response capability vectors, a feasibility screening operation for energy-type energy storage scheduling is performed. The controller pre-stores the qualification thresholds corresponding to three characteristic parameters: the lower limit threshold for available energy storage capacity, the lower limit threshold for charge-discharge conversion efficiency, and the lower limit threshold for continuous charge-discharge support duration. The controller iterates through each set of low-frequency response capability vectors, comparing the three characteristic values ​​within the vector with the corresponding qualification thresholds one by one. When all three characteristic values ​​of a certain energy storage unit meet the threshold requirements, the energy storage unit is included in the set of energy-type energy storage candidate units; any energy storage unit that does not meet any of the threshold requirements is not included in the set of energy-type energy storage candidate units.

[0065] For all energy storage units within the candidate set of energy-type energy storage units, a scheduling priority evaluation calculation is performed. The controller pre-configures priority weight coefficients for three characteristic parameters. The weight coefficients are set according to the dependence of low-frequency energy scheduling on each characteristic, with the current available energy storage capacity having the highest weight, followed by charge-discharge conversion efficiency, and the continuous charge-discharge support duration having the lowest weight. For example, the weight for the current available energy storage capacity can be configured as 0.5, the weight for charge-discharge conversion efficiency as 0.3, and the weight for continuous charge-discharge support duration as 0.2, with a sum of 1. During the calculation, an extreme value normalization method is used to normalize each characteristic parameter. For positive indicators such as current available energy storage capacity, charge-discharge conversion efficiency, and continuous charge-discharge support duration, the normalized value is the current characteristic value divided by the maximum value of that characteristic within the candidate unit set. The normalized feature values ​​are multiplied by their corresponding weights, and the sum of the three products is used to obtain the priority comprehensive score of a single energy storage unit. After all candidate units have been calculated, they are sorted in descending order of comprehensive score. This sorting rule is the priority condition for energy storage.

[0066] Next, after obtaining the energy storage priority conditions, scheduling and allocation are performed based on the total power demand of the low-frequency energy components. The controller reads the current low-frequency energy component values ​​to determine the total charging and discharging power value and power direction to be allocated. Positive low-frequency energy component values ​​correspond to the power demand in the discharging direction, while negative values ​​correspond to the power demand in the charging direction. Power is allocated sequentially according to the energy storage priority conditions, starting with the highest priority energy storage unit. The highest priority unit is allocated its current stable output maximum charging and discharging power first. If there is still a surplus in the total power demand, power is allocated to the next priority energy storage unit until the power demand of all low-frequency energy components is completely covered. All participating energy storage units and their corresponding target power together constitute the energy storage scheduling scheme.

[0067] Finally, risk prediction feedforward compensation is performed. Specifically, risk prediction adopts a time-series extrapolation calculation method. Based on the current state of charge (SOC) of each energy storage unit and the initially allocated target power, the controller extrapolates the trend of SOC change for each unit over a fixed period of time. The fixed period can be set to 30 minutes, and the extrapolation calculation step size is set to 1 minute. The updated SOC value is calculated every one step. During the extrapolation calculation, a positive allocated power corresponds to the discharge direction, and the SOC decreases at the corresponding rate; a negative allocated power corresponds to the charging direction, and the SOC increases at the corresponding rate. The updated SOC value is calculated based on the ampere-hour integral principle, that is, the current SOC value minus or plus the ratio of the cumulative charge / discharge capacity to the rated energy storage capacity within the current step size. The SOC decreases in the discharge direction and increases in the charging direction. The preset safe upper limit for the SOC is 90%, and the safe lower limit is 20%.

[0068] Taking a single energy storage unit with a rated capacity of 2MWh as an example, its initial state of charge (SOC) is 25%, and the initial allocated discharge power corresponds to an 8% decrease in SOC per hour. Table 1 shows the segmented SOC data before and after feedforward compensation within a 30-minute simulation period. The continuous trend of the two sets of data is as follows: Figure 2 As shown, Figure 2 The 20% state of charge safety lower limit reference line is marked in the middle, which intuitively shows the effect of feedforward compensation on suppressing the rate of decrease in state of charge, and reflects the role of risk prediction feedforward compensation in avoiding the risk of exceeding the operating limit in advance.

[0069] Table 1: Comparison of Energy Storage Unit State of Charge Projection Data

[0070] If the simulation results show that the state of charge (SBC) of a certain energy storage unit will reach the preset safety upper and lower limits within the simulation period, the unit is deemed to have an operational risk. The controller reduces the allocated power of the unit by 30% (example), and the resulting power difference is transferred to subsequent risk-free energy storage units in priority. After the power adjustment is completed, the simulation results of all units are verified again until all units meet the safety boundary requirements. If, after multiple rounds of adjustment, there is still a power difference that cannot be absorbed through transfer, this difference is marked as an energy-type dispatch power deficit and reported to the upper-level management unit of the energy storage system for processing. Finally, the target power, runtime constraints, and other parameters of each energy storage unit are compiled to generate the energy-type dispatch parameter configuration.

[0071] By constructing low-frequency capacity vectors through feature extraction, determining energy-type candidate units through threshold screening, clarifying scheduling priorities through weighted scoring, and then allocating low-frequency energy demands in sequence with corresponding risk feedforward simulation and compensation processing, the system achieves the effect of stably carrying basic energy scheduling needs while avoiding operational overrun risks in advance and improving the operational safety of energy-type energy storage.

[0072] Furthermore, the method provided in this application embodiment includes: SOC features are extracted based on the state data of each unit to obtain the SOC feature matrix; the SOC feature matrix is ​​then subjected to influence mapping based on the call direction bias component to obtain the SOC change path; an initial correction strategy is generated based on the SOC change path, and the initial correction strategy is optimized for safety boundaries and compensated for high-frequency response concessions to generate the SOC correction parameter configuration.

[0073] Optionally, the energy storage controller extracts four characteristic parameters from the real-time acquired status data of each energy storage unit: the current state of charge (SOC) value, the upper limit of the allowable SOC operation, the lower limit of the allowable SOC operation, and the current available SOC margin for each unit. These four characteristic parameters for a single energy storage unit are arranged in rows, forming a two-dimensional SOC feature matrix. Each row in the matrix corresponds to one energy storage unit, and each column corresponds to one SOC characteristic parameter. All values ​​are directly read from the acquired status data and can be directly used for subsequent mapping analysis.

[0074] After constructing the SOC feature matrix, influence mapping calculations are performed in conjunction with the call direction bias component. The sign of the call direction bias component represents the overall charging and discharging bias within the current window, and the magnitude of the value represents the degree of bias. The mapping calculation adopts a linear mapping method: first, the value of the call direction bias component is divided by the total duration of the current sliding filter window to obtain the direction bias intensity per unit time; then, the change in state of charge (SOC) of each unit per unit time is calculated by combining the rate of change of SOC corresponding to the rated charging and discharging power of each energy storage unit. The rate of change of SOC can be calculated by dividing the rated charging and discharging power by the rated energy storage capacity, and the unit is the percentage change per hour. This rate of change characterizes the percentage change of SOC per hour under the action of unit power, representing the magnitude of the change in SOC per unit time under unit power. Starting from the current SOC value, the continuous change trajectory of SOC within a 15-minute timeframe is extrapolated along the bias direction. This trajectory is the SOC change path of the corresponding energy storage unit, and the change paths of all units together constitute the complete mapping result. The simulation duration can be pre-configured based on the typical issuance cycle of grid dispatch instructions, for example, it can be configured to 15 minutes. This duration represents the adjustment time window that the state of charge (SOC) of the energy storage units needs to be monitored in advance under the current dispatch trend. This mapping process transforms the charging and discharging direction bias at the dispatch instruction level into the trajectory offset of the SOC of each energy storage unit, enabling the SOC correction strategy to make adjustments in advance for the continuous charging and discharging trend indicated by the bias direction.

[0075] Next, based on the SOC change path of each energy storage unit, the controller compares the change path of each unit with the reasonable operating range of the state of charge. If the path shows that the state of charge of a certain unit will continue to approach the upper limit of operation in the subsequent period, the initial correction strategy is to increase the discharge power allocation ratio and decrease the charging power allocation ratio of that unit, guiding the state of charge to fall back to the middle of the range. If the path shows that the state of charge will continue to approach the lower limit of operation, the initial correction strategy is to increase the charging power allocation ratio and decrease the discharge power allocation ratio of that unit, guiding the state of charge to rise back to the middle of the range. The initial correction strategy is based on a preset unit correction step size, which can be set to 1% of the rated capacity of the energy storage unit, to initially determine the direction and basic adjustment amount of the state of charge adjustment for each energy storage unit.

[0076] The initial correction strategy is then optimized for safety boundaries. The controller pre-stores the mandatory safety upper and lower limits of the state of charge (SOC) for each energy storage unit. For example, the mandatory safety upper limit is 95%, the mandatory safety lower limit is 10%, and the normal operating upper and lower limits are 90% and 20%, respectively. The mandatory safety limits are stricter than the normal operating upper and lower limits. The target SOC value of each unit under the initial correction strategy is verified one by one. If the target value exceeds the mandatory safety upper and lower limits, the adjustment is truncated to the safety boundary to ensure that the corrected SOC always remains within the mandatory safety range. If multiple units simultaneously exceed the limits, the unit correction step size for each energy storage unit is reduced proportionally to its rated capacity relative to the total capacity of all exceeding units, ensuring that the overall correction direction is consistent with the influence trend of the bias component and avoiding conflicts between the correction action and the scheduling direction.

[0077] After completing the safety boundary optimization, high-frequency response concession compensation is further implemented. The controller extracts all energy storage units with high-frequency response capabilities from the power-type energy storage candidate unit set. For these units, an additional 5% of the rated capacity is reserved as a state of charge (SOC) margin based on the optimized correction strategy. For discharge direction correction, a margin is reserved upward from the forced safety lower limit of SOC to ensure that the corrected SOC is not lower than the sum of the forced safety lower limit and the reserved margin, avoiding excessively low SOC from affecting the high-frequency discharge response. For charging direction correction, a margin is reserved downward from the forced safety upper limit of SOC to ensure that the corrected SOC is not higher than the difference between the forced safety upper limit and the reserved margin, avoiding excessively high SOC from affecting the high-frequency charging response. The compensation-based correction strategy ensures that the high-frequency response units always have sufficient adjustment margin while meeting the SOC adjustment requirements. After completing the above two optimization processes, the target adjustment amount, adjustment rate, and margin reservation parameters for each unit are compiled to generate the final SOC correction parameter configuration.

[0078] By extracting multi-dimensional state of charge features to construct a feature matrix, combining directional bias component mapping to deduce the state change path, and then combining it with dual optimization of safety boundary verification and high-frequency response margin reservation, the goal is to ensure the safe operation of the equipment and not weaken the high-frequency power response capability while correcting the state of charge level of the energy storage unit.

[0079] Furthermore, the method provided in this application embodiment includes: By integrating the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration, a candidate response scheduling strategy is obtained; scheduling conflict incidents are predicted based on the candidate response scheduling strategy to obtain a scheduling conflict incident chain; reserve margin risk incidents are predicted based on the candidate response scheduling strategy to obtain a reserve margin risk incident chain; global compensation is performed on the candidate response scheduling strategy based on the scheduling conflict incident chain and the reserve margin risk incident chain to generate the high-frequency call response strategy.

[0080] In one embodiment, the energy storage controller superimposes the target power parameters of each energy storage unit in the power-type scheduling parameter configuration and the target power parameters of each energy storage unit in the energy-type scheduling parameter configuration, one by one, to obtain the initial total target power value of each unit. Then, it superimposes the power adjustment amount and margin constraint requirements of each unit in the SOC correction parameter configuration, integrating them to form a complete scheduling scheme that includes the target power for charging and discharging of all energy storage units, the action execution sequence, and the operating constraints. This scheme is the candidate response scheduling strategy. The single execution cycle of the candidate response scheduling strategy can be pre-configured according to the control cycle of the power grid dispatching system, for example, it can be configured to 30 seconds, serving as the basis for subsequent risk prediction and global optimization.

[0081] When performing scheduling conflict prediction, a time-series node verification approach is adopted. First, the single execution cycle of the candidate response scheduling strategy is divided into equally spaced time-series nodes, with a time interval of 1 second, resulting in a total of 30 consecutive time-series nodes throughout the cycle. For each time-series node, all power commands of each energy storage unit at the corresponding time are checked one by one. If the same energy storage unit has both charging and discharging power commands at the same time-series node, it is determined to be a power direction conflict; if the total target power of the same energy storage unit exceeds the current maximum allowable charging and discharging power of that unit, it is determined to be a power amplitude conflict. For each identified conflict node, node information is recorded in a fixed format according to the conflict node's time-series position, conflicting energy storage unit number, conflict type, and power exceeding amplitude. All conflict nodes within the execution cycle are then connected in chronological order to form a complete scheduling conflict chain.

[0082] When performing reserve margin risk event prediction, a margin threshold comparison method is adopted. The controller pre-sets the lower limit of the overall reserve power margin and the lower limit of the reserve capacity margin of the energy storage system. For example, the lower limit of the reserve power margin is set to 10% of the total rated power of all power-type energy storage units, and the lower limit of the reserve capacity margin is set to 5% of the total rated capacity of the energy storage system. The reserve power margin is defined as the sum of the current remaining available adjustable power of all power-type energy storage units, and the reserve capacity margin is defined as the sum of the capacity converted from the current remaining available state of charge margin of all energy storage units. For each time-series node, the system operating status at that moment is deduced by combining the target power of each unit in the candidate response scheduling strategy, and the remaining reserve power margin and remaining reserve capacity margin at the current moment are calculated respectively. If any margin value is lower than the preset lower limit, it is determined to be a reserve margin risk node. For each risk node identified, the node information is recorded in a fixed format according to the risk node's time-series position, risk type, current margin value, and margin gap value. All risk nodes in the execution cycle are arranged in series in chronological order to form a complete reserve margin risk event chain.

[0083] Next, based on the scheduling conflict incident chain and the reserve margin risk incident chain, a global compensation adjustment is performed on the candidate response scheduling strategy. For each conflict node in the scheduling conflict incident chain, the power command timing and amplitude of the energy storage unit are adjusted first to resolve power direction conflicts. Since the energy storage unit mainly carries low-frequency energy components, its timing and amplitude adjustments have a relatively small impact on the instantaneous power response of the system. Prioritizing the adjustment of the energy storage unit can maximize the complete response capability of the high-frequency power component while resolving conflicts. For power amplitude conflicts, the portion exceeding the maximum allowable power is distributed to other idle energy storage units of the same type according to the proportion of remaining available power. For the amplitude exceeding the limit that cannot be resolved by distribution, non-core power demand in the low-frequency energy component carried by the energy storage unit is preferentially reduced, while retaining the complete response capability of the high-frequency power component, ultimately limiting the target power to within the maximum allowable power range of the unit.

[0084] For each risk node in the reserve margin risk incident chain, priority is given to reducing the instantaneous power allocation ratio of energy-type energy storage units, reserving more regulation capacity of power-type energy storage as system backup. Simultaneously, the single-correction magnitude of the SOC correction parameter is reduced proportionally to the reserve margin gap at each risk node. Specifically, if the reserve power margin gap is 20% of the preset lower limit, the single-correction step size of the SOC correction parameter is reduced proportionally by 20%, releasing some capacity margin to supplement the system's backup capability. After completing the compensation adjustments for all nodes, conflict verification and margin verification are re-executed throughout the entire execution cycle. Once it is confirmed that there are no new conflicts or risks, all adjusted operating parameters are integrated to generate the final high-frequency call response strategy.

[0085] By fusing three types of scheduling parameters to generate candidate scheduling strategies, verifying and identifying scheduling conflicts and reserve margin risks at each time-sequence node and constructing corresponding accident chains, and carrying out targeted global compensation adjustments, the system aims to eliminate scheduling execution logic conflicts, ensure the system's reserve adjustment capabilities, and ensure that scheduling strategies can be executed stably and have sufficient safety margins in high-frequency calling scenarios.

[0086] In summary, the energy storage system optimization scheduling method for high-frequency dispatch scenarios provided in this application has the following technical effects: This application decomposes power dispatch commands into high-frequency power components, low-frequency energy components, and dispatch direction bias components, and performs power-type energy storage dispatch, energy-type energy storage dispatch, and SOC correction respectively. After global optimization to resolve dispatch conflicts and supplement reserve margin, a response strategy is generated, making the dispatch response of the energy storage system more accurate and the operation more stable and reliable in high-frequency dispatch scenarios. It achieves the technical effect of optimized dispatch that takes into account response speed, equipment life and operational safety in high-frequency power dispatch scenarios.

[0087] Example 2, as Figure 3 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides an energy storage system optimization scheduling device for high-frequency dispatch scenarios, the device comprising: The high-frequency call index calculation module 1 is used to calculate the high-frequency call index according to the power call command sequence of the energy storage system.

[0088] The multi-component acquisition module 2, if the high-frequency call index meets the predetermined high-frequency call index, performs sliding filtering on the power call instruction sequence to obtain high-frequency power components, low-frequency energy components and call direction bias components.

[0089] The power-type parameter configuration acquisition module 3 is used to acquire the status data of each unit of the energy storage system, combine the high-frequency power component to perform power-type energy storage drive scheduling of the energy storage system, and acquire the power-type scheduling parameter configuration.

[0090] Energy type parameter configuration acquisition module 4 is used to perform energy type energy storage drive scheduling on the energy storage system according to the low frequency energy component and the status data of each unit, and to acquire energy type scheduling parameter configuration.

[0091] The high-frequency call response strategy acquisition module 5 is used to generate SOC correction parameter configuration based on the call direction bias component, and to perform global optimization on the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration and the SOC correction parameter configuration to obtain the high-frequency call response strategy.

[0092] Furthermore, the high-frequency call index calculation module 1 is used to perform the following steps: The system calculates the number of calls, average call interval, and average power change rate of the power call instruction sequence; outputs the ratio of the number of calls to the predetermined number of calls as the first parameter of high-frequency call; outputs the ratio of the predetermined call interval to the average call interval as the second parameter of high-frequency call; outputs the ratio of the average power change rate to the predetermined power change rate as the third parameter of high-frequency call; and performs a weighted calculation on the first parameter, the second parameter, and the third parameter of high-frequency call according to the high-frequency call weight condition to obtain the high-frequency call index.

[0093] Furthermore, the multi-type component acquisition module 2 is used to perform the following steps: The sliding filter window length is determined based on the high-frequency call index; based on the sliding filter window length, the current sliding filter window corresponding to the current scheduling time is determined in the power call instruction sequence, and multiple corresponding power call instructions within the current sliding filter window are extracted; sliding filter calculation is performed based on the target call power of each corresponding power call instruction to obtain the low-frequency energy component; the difference between the current target call power corresponding to the current scheduling time and the low-frequency energy component is calculated to obtain the high-frequency power component; based on the multiple corresponding power call instructions, the cumulative call energy in the discharge direction and the cumulative call energy in the charging direction are calculated; the difference between the cumulative call energy in the discharge direction and the cumulative call energy in the charging direction is calculated to obtain the call direction bias component.

[0094] Furthermore, the power parameter configuration acquisition module 3 is used to perform the following steps: Based on the status data of each unit, high-frequency response capability features are extracted from each energy storage unit in the energy storage system to obtain multiple high-frequency response capability vectors. Based on these vectors, power-type energy storage scheduling feasibility is screened for each energy storage unit to obtain a set of power-type energy storage candidate units. Scheduling priority is evaluated on the set of power-type energy storage candidate units to obtain power-type energy storage priority conditions. Based on these priority conditions, scheduling decisions are made on the set of power-type energy storage candidate units according to the high-frequency power components to obtain a power-type energy storage scheduling scheme. Finally, the power-type energy storage scheduling scheme is subjected to amplitude limiting verification and lifetime loss compensation to generate the power-type scheduling parameter configuration.

[0095] Furthermore, the energy-type parameter configuration acquisition module 4 is used to perform the following steps: Based on the status data of each unit, low-frequency response capability features are extracted from each energy storage unit in the energy storage system to obtain multiple low-frequency response capability vectors. Based on these multiple low-frequency response capability vectors, energy storage scheduling feasibility is screened for each energy storage unit to obtain a set of energy storage candidate units. Scheduling priority is evaluated on the set of energy storage candidate units to obtain energy storage priority conditions. Based on these energy storage priority conditions, scheduling decisions are made on the set of energy storage candidate units according to the low-frequency energy components to obtain an energy storage scheduling scheme. Risk prediction feedforward compensation is applied to the energy storage scheduling scheme to generate the energy storage scheduling parameter configuration.

[0096] Furthermore, the high-frequency call response strategy acquisition module 5 is used to perform the following steps: SOC features are extracted based on the state data of each unit to obtain the SOC feature matrix; the SOC feature matrix is ​​then subjected to influence mapping based on the call direction bias component to obtain the SOC change path; an initial correction strategy is generated based on the SOC change path, and the initial correction strategy is optimized for safety boundaries and compensated for high-frequency response concessions to generate the SOC correction parameter configuration.

[0097] Furthermore, the high-frequency call response strategy acquisition module 5 is used to perform the following steps: By integrating the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration, a candidate response scheduling strategy is obtained; scheduling conflict incidents are predicted based on the candidate response scheduling strategy to obtain a scheduling conflict incident chain; reserve margin risk incidents are predicted based on the candidate response scheduling strategy to obtain a reserve margin risk incident chain; global compensation is performed on the candidate response scheduling strategy based on the scheduling conflict incident chain and the reserve margin risk incident chain to generate the high-frequency call response strategy.

[0098] Furthermore, the high-frequency call index calculation module 1 is used to perform the following steps: The high-frequency call weighting conditions include a first high-frequency call weight, a second high-frequency call weight, and a third high-frequency call weight.

[0099] Furthermore, the multi-type component acquisition module 2 is used to perform the following steps: The length of the sliding filter window is negatively correlated with the high-frequency call index.

[0100] The energy storage system optimization scheduling device for high-frequency dispatch scenarios provided in the embodiments of the present invention can execute the energy storage system optimization scheduling method for high-frequency dispatch scenarios provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0101] Although this application makes various references to certain modules in the apparatus according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not intended to limit the scope of protection of this invention.

[0102] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. An optimized scheduling method for energy storage systems in high-frequency dispatch scenarios, characterized in that, The method includes: Calculate the high-frequency dispatch index based on the power dispatch command sequence of the energy storage system; If the high-frequency call index meets the predetermined high-frequency call index, the power call instruction sequence is subjected to sliding filtering to obtain a high-frequency power component, a low-frequency energy component, and a call direction bias component. This includes: determining the sliding filter window length based on the high-frequency call index; determining the current sliding filter window corresponding to the current scheduling time in the power call instruction sequence based on the sliding filter window length; extracting multiple corresponding power call instructions within the current sliding filter window; performing sliding filtering calculation based on the target call power of each corresponding power call instruction to obtain the low-frequency energy component; calculating the difference between the current target call power corresponding to the current scheduling time and the low-frequency energy component to obtain the high-frequency power component; calculating the cumulative call energy in the discharge direction and the cumulative call energy in the charging direction based on the multiple corresponding power call instructions; and calculating the difference between the cumulative call energy in the discharge direction and the cumulative call energy in the charging direction to obtain the call direction bias component. Acquire the status data of each unit of the energy storage system, combine the high-frequency power component to perform power-type energy storage drive scheduling of the energy storage system, and obtain the power-type scheduling parameter configuration; Based on the low-frequency energy component and the state data of each unit, the energy storage system is subjected to energy-type energy storage drive scheduling to obtain energy-type scheduling parameter configuration; Based on the call direction bias component, a SOC correction parameter configuration is generated, and the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration are globally optimized to obtain a high-frequency call response strategy; The process of generating SOC correction parameter configuration based on the call direction bias component includes: SOC features are extracted based on the state data of each unit to obtain the SOC feature matrix; The SOC feature matrix is ​​subjected to influence mapping based on the call direction bias component to obtain the SOC change path; An initial correction strategy is generated based on the SOC change path, and the initial correction strategy is optimized for safety boundaries and compensated for high-frequency response concessions to generate the SOC correction parameter configuration.

2. The energy storage system optimization scheduling method for high-frequency dispatch scenarios as described in claim 1, characterized in that, Based on the power dispatch command sequence of the energy storage system, calculate the high-frequency dispatch index, including: The number of times the power call instruction sequence was called, the average call interval, and the average power change rate were statistically analyzed. The ratio of the number of calls to the predetermined number of calls is output as the first parameter for high-frequency calls; The ratio of the predetermined call interval to the average call interval is output as the second parameter for high-frequency calls; The ratio of the average power change rate to the predetermined power change rate is output as the third parameter for high-frequency calling; The high-frequency call index is obtained by weighting the first parameter, the second parameter, and the third parameter of the high-frequency call according to the high-frequency call weight condition.

3. The energy storage system optimization scheduling method for high-frequency dispatch scenarios as described in claim 1, characterized in that, Obtain the status data of each unit of the energy storage system, combine the high-frequency power component to perform power-type energy storage drive scheduling of the energy storage system, and obtain the power-type scheduling parameter configuration, including: Based on the status data of each unit, high-frequency response capability features of each energy storage unit in the energy storage system are extracted to obtain multiple high-frequency response capability vectors. Based on the multiple high-frequency response capability vectors, the feasibility of power-type energy storage scheduling is screened for each energy storage unit to obtain a set of power-type energy storage candidate units; The scheduling priority of the power-type energy storage candidate unit set is evaluated to obtain the power-type energy storage priority conditions; Based on the power-type energy storage priority conditions, scheduling decisions are made on the power-type energy storage candidate unit set according to the high-frequency power components to obtain a power-type energy storage scheduling scheme. The power-type energy storage scheduling scheme is subjected to amplitude limiting verification and lifetime loss compensation to generate the power-type scheduling parameter configuration.

4. The energy storage system optimization scheduling method for high-frequency dispatch scenarios as described in claim 1, characterized in that, Based on the low-frequency energy component and the state data of each unit, the energy storage system is subjected to energy-type energy storage drive scheduling to obtain energy-type scheduling parameter configuration, including: Based on the status data of each unit, low-frequency response capability features of each energy storage unit in the energy storage system are extracted to obtain multiple low-frequency response capability vectors. Based on the multiple low-frequency response capability vectors, the feasibility of energy storage scheduling for each energy storage unit is screened to obtain a set of energy storage candidate units. The scheduling priority of the set of energy storage candidate units is evaluated to obtain the energy storage priority conditions; Based on the energy storage priority conditions, scheduling decisions are made on the energy storage candidate unit set according to the low-frequency energy components to obtain an energy storage scheduling scheme. Risk prediction and feedforward compensation are performed on the energy storage scheduling scheme to generate the energy scheduling parameter configuration.

5. The energy storage system optimization scheduling method for high-frequency dispatch scenarios as described in claim 1, characterized in that, Global optimization is performed on the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration to obtain a high-frequency call response strategy, including: By integrating the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration, and the SOC correction parameter configuration, a candidate response scheduling strategy is obtained. Based on the candidate response scheduling strategy, predict scheduling conflict incidents and obtain scheduling conflict incident chains. Based on the candidate response scheduling strategy, reserve margin risk events are predicted, and reserve margin risk event chains are obtained. The candidate response scheduling strategy is globally compensated based on the scheduling conflict incident chain and the backup margin risk incident chain to generate the high-frequency call response strategy.

6. The energy storage system optimization scheduling method for high-frequency dispatch scenarios as described in claim 2, characterized in that, The high-frequency call weighting conditions include a first high-frequency call weight, a second high-frequency call weight, and a third high-frequency call weight.

7. The energy storage system optimization scheduling method for high-frequency dispatch scenarios as described in claim 1, characterized in that, The length of the sliding filter window is negatively correlated with the high-frequency call index.

8. An energy storage system optimization scheduling device for high-frequency dispatch scenarios, characterized in that, The apparatus for implementing the energy storage system optimization scheduling method for high-frequency dispatch scenarios according to any one of claims 1-7 includes: The high-frequency call index calculation module is used to calculate the high-frequency call index based on the power call command sequence of the energy storage system. The multi-component acquisition module performs sliding filtering on the power call instruction sequence if the high-frequency call index meets the predetermined high-frequency call index to obtain high-frequency power components, low-frequency energy components and call direction offset components. The power-type parameter configuration acquisition module is used to acquire the status data of each unit of the energy storage system, combine the high-frequency power component to perform power-type energy storage drive scheduling of the energy storage system, and acquire the power-type scheduling parameter configuration. An energy-type parameter configuration acquisition module is used to perform energy-type energy storage drive scheduling on the energy storage system based on the low-frequency energy component and the status data of each unit, and to acquire energy-type scheduling parameter configuration. The high-frequency call response strategy acquisition module is used to generate SOC correction parameter configuration based on the call direction bias component, and to perform global optimization on the power-type scheduling parameter configuration, the energy-type scheduling parameter configuration and the SOC correction parameter configuration to obtain the high-frequency call response strategy.

Images