Capacity optimization method, system and device for multi-time-scale hybrid energy storage system

Abstract

The invention relates to the technical field of power system energy storage and automatic control, and discloses a capacity optimization method, system and device for a multi-time-scale hybrid energy storage system, and the method comprises the steps: building and solving an initial capacity optimization model based on a multi-time-scale decomposition result of new energy power fluctuation, obtaining an initial capacity configuration parameter and a multi-time scale division parameter; controlling the hybrid energy storage system to participate in power grid frequency modulation based on the initial parameters, and collecting operation performance and equipment state data; generating a performance feedback index according to the data, dynamically correcting an adjustable parameter of the capacity optimization model based on the index, re-solving the model, and processing a division parameter and a capacity configuration parameter as associated variables in the process to obtain an updated parameter; and applying the updated parameters to the system and returning to the frequency modulation operation step to form closed-loop optimization. According to the invention, collaborative dynamic optimization of the energy storage capacity and the labor division strategy is realized, the system economy is improved, the service life of equipment is prolonged, and the operation robustness is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system energy storage and automatic control, and particularly relates to a capacity optimization method, system and device of a multi-time scale hybrid energy storage system. BACKGROUND

[0002] With the increasing penetration of new energy represented by wind power and photovoltaic power in the power system, the randomness and volatility of the output of new energy have brought severe challenges to the frequency stability of the power grid. In order to suppress the power fluctuation of new energy and support the frequency of the power grid, a hybrid energy storage system integrates energy storage units with complementary dynamic response characteristics (such as power-type energy storage such as flywheel and super capacitor and energy-type energy storage such as compressed air energy storage and battery) to utilize the regulation advantages of the energy storage units in different time scales, and has become an important technical direction for improving the flexibility and stability of the power grid.

[0003] The technical development in this field mainly experienced the stages from single energy storage configuration to multi-type energy storage combination, from fixed capacity planning to preliminary scenario adaptability design, and from open-loop control to multi-time scale coordinated control. Early research mainly focused on the capacity optimization of single type energy storage, which was difficult to cope with the wide frequency domain fluctuation demand. Subsequently, researchers began to explore hybrid systems containing power-type and energy-type energy storage, and allocated fluctuating power according to frequency bands through multi-time scale decomposition technology, and initially realized the division and cooperation of different energy storage media in fast and slow frequency bands. However, most of the existing methods still adopt the mode of "one-time design and static operation", that is, after determining the energy storage capacity and fixed division strategy based on historical or typical data in the planning stage, the capacity is not adjusted in the long-term operation. This mode has significant limitations: first, the static design based on limited scenarios is difficult to adapt to the continuous dynamic changes of new energy output and power grid frequency modulation demand, which may cause the capacity configuration to deviate from the actual demand, resulting in waste of investment or insufficient regulation capacity; second, the fixed division strategy cannot be adaptively adjusted according to the actual operating state (such as state of charge and equipment health) of the energy storage unit, which easily leads to long-term overload operation of part of the unit and accelerates the wear and tear, or the division ratio is out of adjustment, affecting the overall regulation performance; third, there is a lack of a closed-loop mechanism that feeds back real-time operation data to the capacity and strategy optimization link, so that the system cannot continuously learn and improve itself in operation, limiting the economy, reliability and robustness in dealing with extreme disturbances in long-term operation.

[0004] Therefore, how to construct a hybrid energy storage system that can dynamically optimize its capacity configuration and multi-time scale division strategy according to the actual operation performance has become a key problem that needs to be broken through in current technical development. SUMMARY

[0005] To solve the above technical problems, the application provides a capacity optimization method, system and device of a multi-time scale hybrid energy storage system, which are used to improve the accuracy of resource configuration and the adaptability of the system to dynamic environment.

[0006] In a first aspect, the application provides a capacity optimization method of a multi-time scale hybrid energy storage system, which comprises the following steps: Step S1: based on the multi-time scale decomposition result of new energy power fluctuation, an initial capacity optimization model is established and solved to obtain initial capacity configuration parameters and initial multi-time scale division parameters of the hybrid energy storage system, wherein the hybrid energy storage system comprises a first type of energy storage unit and a second type of energy storage unit which are complementary in dynamic response characteristics; Step S2: based on the initial capacity configuration parameters and the initial multi-time scale division parameters, the hybrid energy storage system is controlled to participate in grid frequency modulation, and performance data and equipment state data in the running process are collected; Step S3: based on the performance data and equipment state data, performance feedback indicators are generated, the adjustable parameters of the capacity optimization model are dynamically modified according to the performance feedback indicators, and the modified capacity optimization model is re-solved, wherein in the re-solution process, the multi-time scale division parameters and the capacity configuration parameters are treated as associated variables, and updated capacity configuration parameters and updated multi-time scale division parameters of the next running period are obtained; Step S4: the updated capacity configuration parameters and the updated multi-time scale division parameters are applied to the hybrid energy storage system, and the step S2 is returned to form a closed-loop optimization process.

[0007] In a second aspect, the application provides a capacity optimization system of a multi-time scale hybrid energy storage system, which comprises the following modules: An initial configuration module is configured to establish and solve an initial capacity optimization model based on the multi-time scale decomposition result of new energy power fluctuation to obtain initial capacity configuration parameters and initial multi-time scale division parameters of the hybrid energy storage system, wherein the hybrid energy storage system comprises a first type of energy storage unit and a second type of energy storage unit which are complementary in dynamic response characteristics; A frequency modulation running and data collection module is configured to control the hybrid energy storage system to participate in grid frequency modulation based on the initial capacity configuration parameters and the initial multi-time scale division parameters, and collect performance data and equipment state data in the running process; a closed-loop feedback and reconfiguration module configured to generate performance feedback indicators based on the performance data and the equipment state data, and dynamically modify adjustable parameters of the capacity optimization model according to the performance feedback indicators, and re-solve the modified capacity optimization model, wherein, in the re-solution process, the multi-time-scale division parameters and the capacity configuration parameters are treated as associated variables, and updated capacity configuration parameters and updated multi-time-scale division parameters of a next operation period are obtained; an application and iteration module configured to apply the updated capacity configuration parameters and the updated multi-time-scale division parameters to the hybrid energy storage system, and return to execute the frequency modulation operation and data acquisition module, forming a closed-loop optimization process.

[0008] The third aspect of the present application provides a capacity optimization device for a multi-time-scale hybrid energy storage system, comprising a memory and at least one processor, the memory storing instructions; the at least one processor calling the instructions in the memory to enable the capacity optimization device for the multi-time-scale hybrid energy storage system to perform the capacity optimization method for the multi-time-scale hybrid energy storage system as described above.

[0009] Compared with the prior art, the present application has the following advantages: 1. In the technical solution provided by the present application, a closed-loop feedback and reconfiguration mechanism is constructed to convert real-time frequency modulation performance data and equipment state data into a basis for modifying the capacity optimization model, so that the capacity configuration parameters and the division parameters can be updated periodically or based on event triggering, breaking the limitations of traditional static planning, enabling the "hardware capacity" and "task allocation strategy" of the energy storage system to adapt to the actual operation conditions of the power grid and the fluctuation characteristics of new energy, and significantly improving the accuracy of resource allocation and the adaptability of the system to dynamic environments.

[0010] 2. The present application introduces performance feedback indicators (such as frequency deviation and equipment wear times) and associates them to the objective function and safety constraints of the optimization model, automatically balancing frequency modulation performance, investment cost and equipment life loss during the optimization process. The optimization result can guide the system to avoid overload operation and deep cycling of specific energy storage units, such as reducing frequent deep charging and discharging of flywheel energy storage, and maintaining the pressure of compressed air energy storage within a safer range, thereby effectively reducing the operating stress and life loss of key equipment, prolonging the overall service life of the system, and optimizing the life cycle cost while ensuring frequency modulation performance.

[0011] 3.The application integrates intelligent triggering, update verification, performance root cause analysis and other modules, so that the system has the ability of self-sensing, self-diagnosis, self-decision and self-optimization, can flexibly start optimization according to preset conditions or external instructions, safely verify the new configuration, accurately locate the source of performance deviation through root cause analysis, and implement directional correction; such intelligent closed-loop architecture not only improves the robustness of the system in response to complex working conditions and extreme disturbances, but also ensures the smoothness and safety of the configuration update process, and finally enables the hybrid energy storage system to operate stably as a highly reliable, adaptive and sustainable evolution intelligent frequency modulation resource. BRIEF DESCRIPTION OF DRAWINGS

[0012] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced as follows. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor based on these drawings.

[0013] Figure 1 The step flow chart of the capacity optimization method of the multi-time scale hybrid energy storage system in the embodiment of the present application is shown in the figure. Figure 2 The multi-time scale decomposition result of the wind farm power in the embodiment of the present application is shown in the figure. Figure 3 The power distribution and safety redistribution of the flywheel energy storage and compressed air energy storage in the embodiment of the present application is shown in the figure. Figure 4 The frequency modulation tracking error distribution in the embodiment of the present application is shown in the figure. Figure 5 The composition structure diagram of the capacity optimization system of the multi-time scale hybrid energy storage system in the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0014] The embodiment of the present application provides a capacity optimization method, system and device of a multi-time scale hybrid energy storage system. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the term "includes" or "has" and any variation thereof is intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not necessarily limit to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0015] Embodiment one: For ease of understanding, the specific process of the embodiments of the present application is described below. Please refer to Figure 1 The capacity optimization method of a multi-time scale hybrid energy storage system in the embodiments of the present application includes: Step S1: based on the multi-time scale decomposition result of new energy power fluctuation, an initial capacity optimization model is established and solved to obtain initial capacity configuration parameters and initial multi-time scale division parameters of the hybrid energy storage system, wherein the hybrid energy storage system contains a first type of energy storage unit and a second type of energy storage unit with complementary dynamic response characteristics.

[0016] Among them, step S1 further includes: the first type of energy storage unit is flywheel energy storage or super capacitor energy storage; the second type of energy storage unit is compressed air energy storage, flow battery or hydrogen energy storage; the multi-time scale decomposition result includes high frequency component, medium frequency component and low frequency component; the multi-time scale division parameter is used to define the power allocation ratio of the first type of energy storage unit and the second type of energy storage unit to different time scale power components; the solution result of the capacity optimization model is configured: the high frequency component is mainly borne by the first type of energy storage unit, the low frequency component is mainly borne by the second type of energy storage unit, and the medium frequency component is borne by the first type of energy storage unit and the second type of energy storage unit according to the multi-time scale division parameter.

[0017] Specifically, the present specific embodiment aims to explain in detail how to complete the initial capacity and division strategy configuration of the hybrid energy storage system to solve the technical problems in the prior art that the energy storage system planning is disconnected with the actual multi-time scale power fluctuation characteristics of the power grid and cannot adapt to future operation requirements due to relying on one-time static design.

[0018] Specifically, in one specific embodiment of the present application, the initial configuration work begins with the preprocessing and analysis of the historical output data of the target new energy station, such as a 100-megawatt wind farm: first, the empirical mode decomposition method is used to process the fluctuation sequence of the net output power of the station, the principle of this method is to adaptively decompose the non-stationary and nonlinear power signal into a series of intrinsic mode function components arranged from high frequency to low frequency and a residual trend item; in this application scenario, the decomposition process effectively separates the second-level fast disturbance, minute-level medium fluctuation and ten-minute-level slow trend contained in the power fluctuation; specifically, the first intrinsic mode function component usually corresponds to the high-frequency component, with a time scale of 1 second to 10 seconds, mainly reflecting the instantaneous power impact caused by wind speed turbulence or rapid movement of cloud layer; the subsequent several components jointly constitute the medium-frequency component, with a time scale extending to 10 seconds to 5 minutes, embodying the medium-term fluctuation characteristics of power; and the final residual trend item represents the low-frequency component, with a time scale of more than 5 minutes, corresponding to the long-term output change trend caused by weather system changes; through this decomposition, the distribution characteristics of the grid frequency modulation demand in the frequency domain are clearly quantified, providing a direct input basis for subsequent allocation of energy storage adjustment tasks according to different time scales.

[0019] To more intuitively show the multi-time scale decomposition of the actual energy storage state signal, Figure 2 The wind farm power multi-time scale decomposition result diagram is shown, and the fluctuation sequence in a typical operation cycle is decomposed in this embodiment, which directly reveals the distribution characteristics of the pressure fluctuation in different time scales, further verifying the effectiveness of the multi-time scale decomposition method in the state analysis and task allocation of the hybrid energy storage system.

[0020] Based on the time series of power components of each frequency band obtained by the above decomposition, an initial capacity optimization model is constructed; the capacity optimization model is a multi-objective constraint optimization problem considering technical feasibility, economy and reliability, and its core principle is to seek the optimal balance point between the system life cycle cost and the frequency support performance under a series of physical and engineering hard constraints; the solution target of the capacity optimization model directly corresponds to the initial configuration parameters to be determined, mainly including: the rated power and rated energy storage capacity of the first type of energy storage unit such as flywheel energy storage, the rated power and rated energy storage capacity of the second type of energy storage unit such as compressed air energy storage, and a set of multi-time scale division parameters, which are used to quantitatively define the power bearing proportion of the two types of energy storage units in different frequency bands, for example, set a parameter α to represent the power proportion borne by the flywheel energy storage in a certain medium frequency band, then the proportion borne by the compressed air energy storage is 1-α, for the high frequency band and the low frequency band obtained by decomposition, the division relationship can be pre-set according to the energy storage characteristics, for example, the high frequency fluctuation is mainly responded by the flywheel energy storage, and the low frequency fluctuation is mainly smoothed by the compressed air energy storage.

[0021] The objective function of the capacity optimization model is usually set to minimize the equivalent annual cost of the system, which includes the investment cost, the estimated operation and maintenance cost and the depreciation cost calculated by the expected cycle life depreciation; at the same time, in order to ensure the basic frequency modulation efficiency, a penalty term is introduced in the objective function to minimize the cumulative deviation energy between the actual output of the energy storage and the demand power in each frequency band within the evaluation period; the constraint condition system of the capacity optimization model is the key to its close combination with the engineering practice of power system, mainly including: first, the power balance constraint, which ensures that the total adjustment output of the energy storage for each frequency band can cover the power fluctuation demand of the frequency band at each calculation time; second, the physical and safe operation constraints of the energy storage unit itself, for example, the instantaneous output of the flywheel energy storage cannot exceed the power limit of its power electronic converter, and the stored energy must be maintained between the minimum energy corresponding to the safe speed and the rated capacity, the compressed air energy storage needs to meet the power limit of its compressor / expander and the pressure safety upper and lower limit of the gas tank, and the pressure limit is directly related to the available energy boundary through the gas state equation; third, the range constraint of the division parameter; fourth, the systematic performance constraint of meeting the grid connection technical specification requirements, such as the response time, regulation rate and steady-state frequency deviation allowed range of primary frequency modulation, etc.

[0022] After the objective function and the constraint condition are determined, the embodiment adopts a mature optimization algorithm for numerical solution; for example, a non-dominated sorting genetic algorithm with an elitist strategy can be used to process a multi-objective optimization problem to obtain a set of non-worsening Pareto optimal solutions for a decision maker to select according to preferences; or after the weight coefficients of the sub-targets are determined, a nonlinear programming algorithm such as sequential quadratic programming is used to obtain a single optimal solution; the solution process needs to input specific cost parameters such as unit power cost, unit capacity cost, device efficiency parameters and life model parameters, and the power demand time series of each frequency band obtained by the empirical mode decomposition above as the core input data; after the solution is completed, the algorithm outputs specific numerical results, that is, the initial operation configuration scheme of the system; for example, for the 100-megawatt wind farm described above, the initial configuration obtained by the solution can be: flywheel energy storage rated power 20 megawatts, available energy 1 megawatt-hour, compressed air energy storage rated power 15 megawatts, energy storage 30 megawatt-hours, and the medium-frequency fluctuation component is cooperatively borne by the flywheel and the compressed air according to a pre-calculated ratio such as 70% and 30%; this configuration scheme clearly defines the responsibility division and capability boundary of each type of energy storage in the initial state of the system from the quantitative level, and lays a scientific and executable starting point for subsequent real-time frequency modulation operation and feedback-based closed-loop capacity optimization; the entire modeling and solution process deeply integrates mathematical optimization theory with the physical mechanism of power system frequency modulation, the characteristics of energy storage devices and the economic requirements of projects, ensuring the technical rationality and engineering implementability of the initial configuration scheme.

[0023] Step S2: Based on the initial capacity configuration parameters and the initial multi-time scale division parameters, control the hybrid energy storage system to participate in grid frequency modulation, and collect performance data and device state data during operation.

[0024] The control of the hybrid energy storage system to participate in grid frequency modulation adopts a hierarchical coordination control strategy, which includes: the upper coordinator generates total frequency modulation power instructions and power distribution instructions for each energy storage unit according to the grid frequency deviation, the frequency change rate, and the real-time state margin of the first type of energy storage unit and the second type of energy storage unit; the lower executor performs fast tracking control on the power distribution instructions; when the upper coordinator determines that the real-time state margin of any type of energy storage unit is lower than the safety threshold, the power instruction redistribution logic is started, and the part of the power instruction that the energy storage unit with a real-time state margin lower than the safety threshold cannot safely bear is redistributed to another type of energy storage unit or a backup resource.

[0025] Specifically, the specific embodiment aims to explain in detail how to control the hybrid energy storage system to safely and efficiently perform the grid frequency modulation task based on the initial configuration, and collect key operation data, to solve the technical problem in the prior art that the hybrid energy storage operation control and device safety state are disconnected, lacking adaptive protection and redistribution mechanism, thereby affecting the long-term reliable operation of the system and the effectiveness of the data.

[0026] Specifically, in one specific embodiment, when the system obtains the initial capacity configuration parameters and the multi-time scale division parameters, it immediately enters the frequency regulation operation and data collection stage, the core of which is to implement a hierarchical coordinated control strategy; the upper layer coordinator in the strategy is essentially a real-time decision algorithm module running at the millisecond level. The real-time decision algorithm module continuously receives real-time measurement signals from the power grid, mainly including the frequency value f(t) of the grid-connected point and its first derivative df / dt, i.e., the frequency change rate. At the same time, it obtains the real-time state margin of the first type of energy storage unit, such as the flywheel energy storage FESS, and the second type of energy storage unit, such as the compressed air energy storage CAES, through the internal communication bus. For flywheel energy storage, its state margin is usually represented by the state of charge or the equivalent speed percentage, defined as the ratio of the current available energy to the maximum available energy. Therefore, the multi-time scale division + safety protection control logic proposed in this application can effectively constrain the operating range of flywheel energy storage during frequency regulation operation, avoiding deep cycling and out-of-range risks, thereby ensuring the safety and reliability of equipment operation.

[0027] For compressed air energy storage, its state margin is mainly determined by the real-time pressure value of the gas tank relative to the upper and lower limits of the safe operating pressure, which can be quantified as a pressure margin coefficient between 0 and 1. The upper layer coordinator first calculates the total frequency regulation power instruction P_ref_total(t) required by the system to support the grid frequency according to the frequency deviation Δf, such as the difference between the current frequency and the rated frequency 50.00 Hz, and the frequency change rate, through a unified algorithm model combining virtual inertia and droop characteristics. The principle of the unified algorithm model is to simulate the frequency response characteristics of synchronous generators, and its output can provide both fast inertia support and steady-state frequency recovery power.

[0028] Subsequently, the coordinator decomposes the total command according to preset multi-timescale division parameters: specifically, it internally maintains a digital filter bank corresponding to the decomposed frequency band in step S1, decomposing P_ref_total(t) into high-frequency, mid-frequency, and low-frequency components in real time; according to the division parameters, the high-frequency components are all directed to flywheel energy storage, the low-frequency components are all directed to compressed air energy storage, and the mid-frequency components are proportionally assigned, for example, 70% to flywheel energy storage and 30% to compressed air energy storage, thereby generating the preliminary power allocation command P_FESS_pre(t) for each energy storage unit. And P_CAES_pre(t); however, this is only an ideal allocation. Before the command is issued, the coordinator will check this preliminary command against the real-time state margin of each energy storage unit. The safety threshold is a pre-set boundary value. For example, the lower limit of safe operation of flywheel energy storage SOC is set to 20%, and the lower limit of safe pressure of compressed air energy storage is a specific value. The coordinator's algorithm includes a real-time logic judgment unit. If it detects that the state margin of a certain energy storage unit is lower than its safety threshold, or if it predicts that its state will soon exceed the limit according to the preliminary command, the power command reallocation logic is immediately started. The core of this logic is to recalculate the power allocation. For example, when the flywheel SOC is close to the 20% lower limit, the algorithm will reduce the medium frequency and even part of the high frequency power command originally allocated to it at a preset smooth rate, and dynamically add this reduction amount ΔP(t) to the command of compressed air energy storage. If compressed air energy storage also has no margin, the excess part can be marked as "unmet power" and the standby resources can be triggered. This redistribution process uses a slope limiter to ensure smooth power command changes and avoid secondary impacts on the power grid; to visually demonstrate the effects of the above power allocation and safe redistribution process, such as Figure 3 The embodiment shown records flywheel energy storage over a typical time period. Compressed air energy storage By comparing the changes in power commands, it can be seen that at a specific moment, such as when the flywheel energy storage margin is insufficient, the system automatically triggers power redistribution, safely and smoothly transferring part of the power commands from the flywheel energy storage to the compressed air energy storage, ensuring the overall operational safety of the system and the continuous execution of the frequency regulation task; after safety verification and redistribution, the final power commands P_FESS_cmd(t) and P_CAES_cmd(t) are sent to the lower-level actuators.

[0029] The lower-level actuators directly drive the power conversion devices of each energy storage unit, such as converters and compressors / expanders. They typically employ a dual-closed-loop control structure based on proportional-integral regulation. The outer loop receives power commands from the upper-level coordinator, while the inner loop performs rapid current tracking control to ensure accurate and fast execution of power commands, with response times down to the millisecond level. Throughout operation, the system synchronously and frequently records two types of data: performance data, including frequency deviation Δf, frequency change rate, and the tracking error between the actual total frequency regulation output and the required power; and equipment status data, including real-time SOC and speed of flywheel energy storage, real-time tank pressure and temperature of compressed air energy storage, and records of each power redistribution event triggered due to insufficient state margin and the amount of energy transferred. This detailed operational data is cached and packaged, providing an indispensable data foundation for subsequent performance evaluation and closed-loop optimization.

[0030] Step S3: Based on performance data and equipment status data, generate performance feedback indicators, and dynamically correct the adjustable parameters of the capacity optimization model according to the performance feedback indicators. Then, re-solve the corrected capacity optimization model. In the re-solution process, the multi-timescale division of labor parameters and capacity configuration parameters are treated as correlation variables, and the updated capacity configuration parameters and updated multi-timescale division of labor parameters for the next running cycle are obtained.

[0031] The dynamic correction of the adjustable parameters of the capacity optimization model includes adjusting at least one of the following parameters: the weight coefficients of each sub-objective function related to frequency performance, investment cost, and equipment life loss in the capacity optimization model; the boundary conditions of the operational safety constraints of the first type of energy storage unit and / or the second type of energy storage unit in the capacity optimization model; and the initial or reference values ​​of the multi-timescale division parameters used in the re-solution process.

[0032] Specifically, this embodiment aims to explain in detail how to dynamically correct the capacity optimization model and generate an updated configuration based on the actual effect of frequency regulation operation and equipment status, so as to solve the core technical problem in the prior art that the energy storage capacity configuration is fixed and cannot be adaptively adjusted according to the operating performance and equipment loss, resulting in long-term performance degradation and economic loss of the system.

[0033] In one specific embodiment, after the system completes a predetermined cycle of frequency regulation operation and collects complete performance data and equipment status data, a closed-loop feedback and reconfiguration process is initiated. First, the system performs statistical processing on the collected raw time-series data to generate quantified performance feedback indicators. These performance feedback indicators include: the integral of the absolute value of frequency deviation reflecting the stability of the power grid frequency, for example, accumulating the absolute value of frequency deviation per second to obtain the cumulative value for the whole day; the peak value of the frequency change rate reflecting the severity of the impact on the power grid; the energy of unmet frequency regulation requirements reflecting the completion of the frequency regulation task, i.e., the total energy that needs to be stored but could not be output due to power or energy limitations; and equipment loss indicators reflecting the health of the equipment, such as the number of deep cycles of flywheel energy storage throughout the day, defined as the number of times the state of charge (SOC) crosses a preset depth threshold, such as counting the process of dropping from 80% to 20% and then rising again as one cycle; and the frequency of compressed air energy storage pressure exceeding the limit, defined as the number of times the pressure of the storage tank falls below the safety lower limit threshold. These performance feedback indicators comprehensively characterize the operating effect of the previous cycle from three dimensions: system performance, task completion, and equipment durability.

[0034] Another key and intuitive evaluation indicator for the completion of frequency regulation tasks is the "power point tracking error," which is the real-time deviation between the actual total output of the hybrid energy storage system and the total frequency regulation power demand command calculated by the upper-level coordinator. To analyze the statistical characteristics of this error, this embodiment performs distribution statistics on the tracking error sequence within the selected operating cycle. The results are as follows: Figure 4 The diagram shown illustrates the distribution of frequency modulation (FM) tracking errors. The horizontal axis represents the tracking error, and the vertical axis represents the frequency or probability density of each error value within the statistical period. Figure 4 As shown, the tracking error is mainly concentrated in a very narrow range near zero, for example, ±0.02MW, exhibiting an approximately normal distribution. Furthermore, the mean and standard deviation of the absolute error values ​​are both at extremely low levels. This error distribution characteristic clearly indicates that, under the current configuration, the comprehensive regulation capability of the hybrid energy storage system can closely follow the frequency regulation requirements brought about by wind power fluctuations, achieving high-precision power tracking within the limited rated power and energy boundaries. This error distribution map quantifies the system's tracking capability boundary in a statistical sense and serves as an important visual basis for assessing whether the current capacity configuration is sufficient and whether the division of labor strategy is effective. It also provides direct performance feedback for possible subsequent model corrections.

[0035] Subsequently, based on these feedback indicators, the system dynamically corrects the adjustable parameters of the initial capacity optimization model. This correction process follows preset mapping rules. For example, if statistics show that the number of deep cycles of flywheel energy storage exceeds the designed daily reference value, it is considered that the flywheel is undertaking an excessive transient regulation task, accelerating its lifespan loss. In this case, the system will automatically increase the weight coefficient of the equipment lifespan loss sub-objective function in the capacity optimization model, for example, increasing its weight from the initial 0.2 to 0.3. This means that in the next round of optimization calculations, the algorithm will be more inclined to choose a configuration scheme that reduces the number of flywheel cycles, even if this may slightly increase investment costs or relax frequency performance. Similarly, if the frequency deviation integral value is significantly higher than the allowable value of the power grid assessment, the system will correspondingly increase the weight coefficient of the frequency performance sub-objective function. Another correction method is to adjust the operational safety constraint boundary conditions of the model. For example, if the pressure... The frequent occurrence of pressure approaching the lower limit during operation of compressed air energy storage indicates that the safety margin reserved by the minimum allowable operating pressure set in the current model is insufficient. Therefore, this constraint will be automatically tightened, increasing the lower pressure limit value in the next round of optimization by a safety margin increment, thereby guiding the new configuration scheme to reserve more energy buffer space for compressed air energy storage. Furthermore, the initial values ​​of multi-timescale division of labor parameters can also be used as adjustable parameters. For example, if analysis reveals a large frequency tuning tracking error in the mid-frequency band and frequent deep flywheel circulation, the system can infer that the current mid-frequency division of labor ratio, such as flywheel:compressed air = 70%:30%, may be unreasonable, causing the flywheel to undertake too much mid-term energy regulation task that should be borne by compressed air. Therefore, before the next optimization solution, the initial reference value of the mid-frequency division of labor parameters can be adjusted towards the compressed air side, for example, preset to 60%:40%, providing a better search starting point for the optimization algorithm.

[0036] After correcting the aforementioned adjustable parameters, the system will invoke the optimization solver to re-solve the corrected capacity optimization model. A key improvement in this process is the collaborative optimization of multi-timescale task allocation parameters and capacity configuration parameters as a set of correlated variables. This means that when searching for the optimal solution, the optimization algorithm simultaneously considers the two coupled problems of "how to allocate tasks" and "how much capacity to configure," automatically seeking the optimal combination that matches the task allocation strategy with the equipment's capacity. For example, the algorithm might discover that under new conditions of increased lifetime loss weight and tightened pressure constraints, the optimal solution is no longer simply increasing flywheel capacity, but rather appropriately reducing the flywheel's share in the mid-frequency band while simultaneously increasing the rated energy of compressed air, thereby achieving the optimal balance between overall cost, performance, and lifetime. After the solution is completed, the system outputs the updated capacity configuration parameters and multi-timescale task allocation parameters for the next operating cycle, such as fine-tuning the flywheel's rated power, increasing the rated energy of compressed air, and updating the mid-frequency task allocation ratio, thus completing a full closed-loop optimization iteration and allowing the system configuration to continuously evolve based on actual operating experience.

[0037] The multi-timescale division of labor parameters and capacity configuration parameters are treated as related variables and can be implemented in one of the following ways: Method 1: The capacity configuration parameters and multi-timescale division of labor parameters are used together as decision variables and simultaneously optimized and determined during the re-solution process; Method 2: First, the multi-timescale division of labor parameters are corrected based on performance feedback indicators to obtain the corrected division of labor parameters. Then, the corrected multi-timescale division of labor parameters are used as fixed conditions to solve the simplified model in the capacity optimization model that only contains capacity configuration variables, so as to determine the updated capacity configuration parameters.

[0038] Specifically, this implementation method aims to explain in detail how to couple the multi-timescale division of labor parameters and energy storage capacity configuration parameters during the capacity reconfiguration process, so as to solve the technical problem that the two types of key variables are interdependent and their independent adjustment during the optimization process can easily lead to suboptimal solutions or configuration mismatches, thereby affecting the overall effect of closed-loop optimization.

[0039] In one specific embodiment, after the performance feedback index is generated and the adjustable parameters of the model are corrected, the key step of resolving the problem is to handle the correlation between the multi-timescale division of labor parameters and the capacity configuration parameters.

[0040] This embodiment provides two optional implementation methods: Method 1 is integrated collaborative optimization. This method incorporates capacity configuration parameters, such as rated power and rated energy, and multi-timescale division of labor parameters, such as the power allocation ratio of each frequency band, into an extended joint optimization model, which is solved simultaneously as a set of decision variables. In this joint optimization model, the objective function integrates investment cost, frequency performance penalty, and equipment lifetime loss cost. Its computational depth depends on the values ​​of the decision variables: the magnitude of the frequency performance penalty depends on whether each energy storage unit can meet the power demand within its corresponding capacity parameter limit under the given division of labor parameters; the lifetime loss cost is related to the expected working cycle depth of each energy storage unit under the division of labor strategy. The optimization solver explores all possible combinations of capacity and division of labor in a single calculation to find the solution that optimizes the overall objective. Mathematically, this method guarantees that the updated parameters of the output are globally optimal and self-consistent, that is, the division of labor strategy and hardware capabilities achieve the best match, but the computational complexity is relatively high.

[0041] Method two is a two-step sequential optimization. This method decomposes the complex joint optimization problem into two logically consecutive stages to reduce the difficulty of solving it. In the first stage, the system uses a rule-based or fast evaluation algorithm based on performance feedback indicators to specifically correct the division of labor parameters. For example, if the analysis finds that the flywheel energy storage is taking on too much mid-frequency energy tasks, resulting in frequent deep circulation, the algorithm can adjust the reference value of the proportion of mid-frequency flywheels to be borne downward according to the preset adjustment logic. This yields a set of corrected division of labor parameters. In the second stage, the corrected division of labor parameters determined in the first stage are treated as fixed conditions to construct a simplified capacity optimization model. The decision variables of this simplified model only include the rated power and rated energy of various types of energy storage. Since the division of labor strategy is determined, the calculation of cost terms related to task allocation in the objective function is simplified. The optimization problem is transformed into finding the optimal equipment capacity configuration that satisfies all technical constraints and minimizes total investment and operating costs under a given task allocation scheme. This approach, by first clarifying "how tasks are allocated" and then determining "how much capacity needs to be configured," achieves decoupling and sequential optimization of the two types of parameters, resulting in higher computational efficiency and better aligning with the engineering intuition of "strategy-guided configuration." Regardless of whether method one or method two is used, the core objective is to ensure that the newly generated capacity configuration parameters and the multi-timescale task allocation parameters have inherent logical consistency and physical feasibility, avoiding mismatches such as "high task allocation requirements but insufficient capacity support" or "large capacity redundancy but inefficient task allocation strategy," thereby ensuring the technical rationality and engineering value of the closed-loop optimization results.

[0042] The execution of step S3 is controlled by any of the following conditions: the time control parameter reaches the preset fixed period duration; any performance feedback indicator exceeds the corresponding warning threshold; or an update instruction is received from an external scheduling system.

[0043] Specifically, this embodiment aims to explain in detail how to control the start time of the closed-loop optimization process in order to solve the technical problem that unclear start conditions of the closed-loop update process may lead to waste of optimization computing resources, delayed response to sudden performance degradation, or inability to coordinate with the overall power grid dispatch.

[0044] In one specific embodiment, step S3 is not unconditionally periodic, but controlled by a multi-mode intelligent triggering mechanism. This mechanism is executed by the system's built-in closed-loop triggering management module. The triggering mechanism integrates planning, responsiveness, and coordination. Its core lies in determining when to initiate a complete "data evaluation-model correction-resolution" process. The first triggering condition is based on time control parameters, i.e., a preset fixed cycle length. This is a planned triggering mode. For example, the system can be configured to automatically start closed-loop optimization at midnight every day or midnight every Monday. This mode ensures that the system can systematically review and update the accumulated operating data of the past cycle, such as the previous day or week, at a stable rhythm. It is suitable for scenarios with relatively stable operating environments that need to align with the daily or weekly plans of the power grid. In this mode, regardless of whether there are any unexpected events, the optimization process will start on time, ensuring the basic frequency and regularity of configuration updates.

[0045] The second triggering condition is event-driven, meaning it is triggered immediately when any performance feedback metric exceeds its corresponding warning threshold. The warning threshold is a pre-set critical value based on technical standards, equipment tolerance, or operational experience. The closed-loop trigger management module calculates and monitors these metrics in real-time or near real-time. Once any metric is detected to exceed its threshold, the module immediately generates a high-priority interrupt signal, suspends any other background tasks that may be in progress, and forcibly starts the closed-loop optimization process. This triggering method enables the system to respond quickly to performance degradation or abnormal equipment pressure, allowing for timely configuration adjustments before problems accumulate to a more serious level, demonstrating the agility and self-protection characteristics of the closed-loop system.

[0046] The third trigger condition is receiving an update command from an external dispatch system. This provides an interface for the hybrid energy storage system to coordinate with the upper-level power grid dispatch center for local self-optimization. External commands are usually sent in the form of digital messages through standard power communication protocols. The message content can include command codes and parameters. For example, when the weather forecast indicates that there will be strong winds in the next 24 hours, which may cause drastic fluctuations in wind power, the dispatch center can send a command with the code "OPT_REQUEST" to the hybrid energy storage system. After receiving and parsing the command, the closed-loop trigger management module will immediately start a closed-loop optimization, regardless of whether the fixed cycle has been reached or the indicator has exceeded the limit. This allows the optimization rhythm of the system to conform to the overall operation strategy and risk prevention and control arrangements of the power grid, realizing an effective combination of local adaptation and global centralized dispatch. The above three trigger conditions work in parallel through a logical OR relationship, and usually the event trigger is set with the highest priority, followed by the external command, and the fixed cycle is used as a backup. Together, they constitute a flexible, reliable, and tightly coupled closed-loop start-up control logic with the power grid operation.

[0047] The process, following step S3 and preceding step S4, includes an update verification step: Based on the updated capacity configuration parameters and updated multi-timescale division of labor parameters, a short-term trial run of the hybrid energy storage system is conducted. During this trial run, the system's operational performance indicators and equipment status indicators are monitored and collected. If both the operational performance indicators and equipment status indicators meet the preset access criteria, the updated capacity configuration parameters and updated multi-timescale division of labor parameters are confirmed to be applied to subsequent frequency regulation operations. Otherwise, a rollback to the capacity configuration parameters and multi-timescale division of labor parameters of the previous operating cycle is triggered, or the updated capacity configuration parameters and updated multi-timescale division of labor parameters are limited and corrected before re-verification. The access criteria are a pre-set set of quantitative or qualitative standards used to determine whether the capacity configuration parameters and multi-timescale division of labor parameters of the next operating cycle can be formally put into operation.

[0048] Specifically, this implementation method aims to explain in detail how to perform engineering verification and security access for the new configuration parameters generated by closed-loop optimization, so as to solve the technical problem that directly deploying optimization results may lead to system instability, performance degradation or even equipment security risks due to model errors or sudden changes in the scenario.

[0049] In one specific embodiment, after calculating the updated capacity configuration parameters and multi-timescale division of labor parameters through closed-loop feedback and reconfiguration steps, the system does not immediately apply them to long-term operation. Instead, it first executes an update verification step. The core of this step is to organize a short-term trial run. The system controller loads the new configuration parameters but marks it as a "trial run" state and sets a short observation period, such as the next 30 minutes to 2 hours. During this trial run, the hybrid energy storage system participates in real-time grid frequency regulation based on new parameters such as the new flywheel rated power and the new intermediate frequency division of labor ratio. However, all its operation and monitoring... The testing logic is exactly the same as that in the formal operation phase; the system synchronously monitors and collects two types of key data: one is the operation performance indicators, which mainly refer to the stability of the grid frequency during the trial operation, such as the absolute value of the average frequency deviation, the maximum value of the frequency change rate, and the real-time tracking error of the frequency regulation power during this period; the other is the equipment status indicators, which focus on the operating boundary pressure of the energy storage unit under the new configuration, such as whether the state of charge of flywheel energy storage fluctuates more frequently or more significantly, whether it has reached the new energy limit, and whether the tank pressure of compressed air energy storage approaches its safe pressure lower limit faster under the newly set rated power output.

[0050] After the trial run, the system will compare the collected indicator data with the access criteria stored in the database. The access criteria are a set of quantitative standards that define whether the new configuration can be safely and effectively put into use. The criteria are set from technical procedures, equipment safety specifications and historical operating experience. If all monitoring indicators meet these preset criteria, the verification is successful. The system will confirm that the updated configuration parameters are officially effective and will then proceed to the subsequent long-term frequency regulation operation.

[0051] If the verification fails, meaning any one or more indicators do not meet the admission criteria, the system triggers preset fault-tolerant processing logic. One approach is direct rollback, whereby the new parameters generated by this optimization are discarded, and the system controller automatically switches back to the configuration parameters that have been verified to be stable in the previous operating cycle, ensuring uninterrupted frequency regulation service. Another, more refined approach is to re-verify after limiting correction. For example, if the trial operation finds that the compressed air storage pressure drops too quickly under the new configuration, a temporary limiting coefficient can be automatically applied to the rated power parameter of the compressed air in the new configuration, such as temporarily adjusting the maximum allowable output power to 90% of the original new value, or fine-tuning the intermediate frequency division parameters, such as slightly reducing the proportion of compressed air in the intermediate frequency. Then, based on this corrected parameter set, a new round of trial operation verification, which may be shorter in duration, is initiated. This process can be iterated until a set of parameters is found that both meets the admission criteria and retains as much optimization benefit as possible. Alternatively, a rollback may be determined in the final decision. This update verification step serves as a safety buffer, significantly improving the reliability and robustness of the closed-loop optimization system in practical engineering applications, and avoiding operational risks caused by the mismatch between the theoretical results of the optimization algorithm and the complex field environment. The basis for the amplitude limit correction is derived from the deviation between short-term trial operation monitoring data and the preset "entry criteria". When the new configuration does not fully meet the standards during trial operation, the specific root causes of the deviation are first analyzed, such as the rapid rate of change of key equipment state quantities like pressure or the deterioration of performance indicators like frequency recovery time. Subsequently, the required theoretical correction amount is quickly estimated based on the trial operation data. To ensure that the update process is absolutely stable and safe, all corrections are strictly limited by the preset maximum single correction step size. If the theoretical value exceeds this step size, only the step size value is applied for amplitude limit correction. At the same time, following the fundamental principle of "safety constraints are only tightened and never loosened", a set of slightly adjusted new parameters is generated for re-verification, achieving safe and gradual parameter optimization.

[0052] Before dynamically correcting the adjustable parameters of the capacity optimization model based on performance feedback indicators, the process includes a performance root cause analysis step: This involves fusing performance data and equipment status data to generate a feature vector characterizing the overall state of the hybrid energy storage system. The feature vector is then matched against a pre-defined diagnostic rule base, which defines the mapping relationship between different abnormal state patterns and potential root causes. This rule base is constructed based on historical operating data and simulation data. Based on the matching analysis, a diagnostic conclusion indicating the potential root cause of the current frequency regulation performance deviation is output. The types of diagnostic conclusions include insufficient total frequency regulation capacity configuration of the hybrid energy storage system, unreasonable multi-timescale division of labor between the first and second types of energy storage units, or excessive operating load on any type of energy storage unit. Based on the diagnostic conclusions, a targeted correction strategy corresponding to the type of diagnostic conclusion is selected and executed to correct the adjustable parameters of the capacity optimization model.

[0053] Specifically, this implementation method aims to explain in detail how to identify the root cause of frequency modulation performance problems through intelligent diagnosis and drive targeted correction in closed-loop optimization, so as to solve the defects of blind and inefficient correction actions that may be caused by simply relying on index values ​​for simple feedback adjustment, which cannot accurately solve complex coupled technical problems.

[0054] In one specific embodiment, before directly adjusting the model parameters based on performance feedback indicators, the system adds a performance root cause analysis step. This step begins with a deep fusion analysis of the original operating data. The heterogeneous data collected in step S2, such as frequency deviation, ROCOF sequence, historical state of charge curve and deep cycle count of flywheel energy storage, real-time pressure curve and pressure low limit exceedance times of compressed air energy storage, and unmet frequency regulation power gap records, are time-aligned and normalized. These refined data are combined into a feature vector, which quantitatively characterizes the comprehensive state of the hybrid energy storage system in the past operating cycle from multiple dimensions such as frequency quality, equipment stress, and task completion.

[0055] Subsequently, the feature vector is matched and analyzed against a pre-built diagnostic rule base. This rule base is not a black-box model generated by machine learning, but a set of judgment rules with a clear logical structure constructed from domain expert knowledge. Each rule defines the mapping relationship between a specific abnormal state pattern and one or more potential root causes. For example, a typical rule can be expressed as follows: if the "average frequency deviation" in the feature vector is greater than 0.06 Hz, and the proportion of "unmet frequency modulation energy" to the total required energy exceeds 5%, but the "flywheel average SOC" is in the normal range of 40%-60% and the "minimum compressed air pressure" is higher than the safety lower limit, then the diagnostic conclusion is "insufficient total frequency modulation capacity configuration". Another rule might stipulate that if the "frequency recovery time, such as the time it takes to recover from a drop to 49.8 Hz to above 49.9 Hz" exceeds 10 seconds, and the "number of deep flywheel cycles" increases significantly, while the "proportion of compressed air output to the total demand for low and medium frequencies" is below a certain threshold, then the diagnosis points to "unreasonable division of labor across multiple time scales, with the flywheel excessively undertaking the task of medium-frequency energy." There are also rules for equipment overload, for example: if the "cumulative time when the flywheel SOC is below 20%" exceeds 15% of the total time, or the "number of times the compressed air pressure touches the lower limit" exceeds 2 times per hour, then the diagnosis is "the corresponding energy storage unit is operating under excessive load."

[0056] Based on matching analysis, the system outputs a clear diagnostic conclusion, which is not simply "poor performance," but rather identifies the most likely technical root cause of the problem. Following this conclusion, the system selects and executes a targeted correction strategy from a pre-set strategy library. If the diagnosis is "insufficient total capacity," the correction strategy prioritizes significantly increasing the weight coefficient of the frequency performance objective term in the capacity optimization model, and may simultaneously reduce the weight of the investment cost term, thereby guiding the algorithm to output a larger rated power and energy configuration in the next round of solving. If the diagnosis is "unreasonable division of labor," the correction strategy focuses on adjusting the initial reference values ​​of the multi-timescale division of labor parameters used in the next round of optimization. For example, it may directly reduce the initial value of the proportion of the mid-frequency flywheel's load by 5%-10%, and may trigger a reassessment of the timescale division threshold. If the diagnosis is "excessive load on specific equipment," the correction strategy will focus on tightening the operational safety constraints of that equipment in the optimization model. For example, it may increase the minimum permissible SOC constraint value of the flywheel or the minimum permissible pressure constraint value of compressed air, forcing more safety margins to be reserved for the equipment. Through this intelligent chain of "data fusion → pattern recognition → root cause diagnosis → targeted measures", the system has achieved a closed-loop decision-making upgrade from perceiving phenomena to understanding the essence and then to precise intervention, which greatly improves the effectiveness and engineering practicality of dynamic parameter correction.

[0057] Step S4: Apply the updated capacity configuration parameters and the updated multi-timescale division of labor parameters to the hybrid energy storage system, and return to step S2 to form a closed-loop optimization process.

[0058] Specifically, this implementation method aims to explain in detail how to safely and smoothly deploy the optimized new configuration to the actual system and start a new round of operation, so as to solve the "last mile" problem of the closed-loop optimization process, ensure that the theoretical optimization results can be transformed into sustainable and iterative actual operation, thereby achieving continuous evolution of system performance.

[0059] In one specific embodiment, after the update verification step confirms that the new set of configuration parameters, including the updated capacity configuration parameters and the updated multi-timescale division of labor parameters, meet all admission criteria, the application and iteration step begins. The primary action in this step is parameter deployment. The system controller's configuration management module writes the new parameters into the runtime database and sets them to an active state. Specifically, the deployment includes: issuing the new rated power and new rated energy of the flywheel energy storage to its local controller; issuing the new rated power and new effective volume of the compressed air storage tank, corresponding to the rated energy, to its control system. At the same time, new multi-timescale division parameters, especially the mid-frequency power allocation ratio, such as updating from flywheel:compressed air = 65%:35% to 60%:40%, and the possible adjustment of timescale frequency division thresholds, are configured into the algorithm core of the upper-level coordinator. To ensure a smooth transition, the deployment process usually adopts a gradual loading strategy. For example, for changes in division parameters, the switch is not instantaneous, but rather a smooth transition from the old value to the new value of the control command allocation weight over several minutes to ten minutes through linear interpolation, so as to avoid the impact of sudden power command changes on the power grid and the energy storage equipment itself.

[0060] After the parameters are deployed and take effect, the system returns to the frequency regulation operation and data acquisition steps, i.e., step S2 above. This means that the hybrid energy storage system will be put into a new round of real-time frequency regulation services for the power grid based on a brand-new, optimized "hardware capability definition" and "task allocation strategy". The control behavior at this time is logically consistent with the initial operation phase, but the configuration basis it is based on has been improved through a round of closed-loop optimization. The system will continue its core work: the upper-level coordinator allocates frequency regulation power commands in real time according to the new division of labor ratio, and the lower-level actuator drives the equipment to respond quickly within the newly defined power and energy boundaries. At the same time, the system continuously collects operating performance data and equipment status data under the new configuration, and its monitoring content and frequency are exactly the same as before.

[0061] This process of "deploying new configuration → operating based on new configuration → collecting new data" marks the end of a complete closed loop and the beginning of the next new closed loop. The newly collected operational data will serve as the input for the next round of closed-loop feedback and reconfiguration steps, i.e., step S3, to evaluate the actual effect of the new configuration and may drive further optimization adjustments. For example, the system may find that under the new division of labor, the pressure fluctuation of compressed air energy storage becomes smoother, but the frequency rapid recovery capability is slightly reduced. This new data performance will be recorded and fed back to provide a basis for the next optimization decision. Through such a recurring iterative cycle of "configuration optimization - deployment and operation - effect evaluation - re-optimization", the capacity configuration and coordination strategy of the entire hybrid energy storage system can continuously approach and dynamically adapt to changes in the actual operating environment of the power grid and frequency regulation requirements, thereby enabling the system to have the ability to continuously improve and evolve. Ultimately, it forms a stable, reliable, and continuously improving long-term operating state, fundamentally realizing the transformation of the energy storage system from a static asset to a dynamic intelligent resource.

[0062] Through the coordination of the above steps, this application improves the matching accuracy of capacity configuration and multi-timescale operation strategies of hybrid energy storage systems, the overall economic efficiency of operation, and the reliability and adaptability of long-term service.

[0063] In summary, this application provides an innovative method for capacity optimization of multi-timescale hybrid energy storage systems. Its core lies in constructing a dynamic adaptive architecture integrating the entire process of "initial configuration - real-time operation - performance evaluation - closed-loop optimization." This method first performs initial modeling and optimization based on the frequency domain decomposition results of new energy power to determine the capacity foundation and task allocation framework of the energy storage system. Then, detailed performance and status data are collected during real-time frequency regulation operation, and these data are used to generate quantitative feedback indicators. Furthermore, through intelligent diagnosis and root cause analysis, key parameters of the optimization model are dynamically corrected. During this process, energy storage capacity configuration and multi-timescale task allocation strategies are treated as closely related variables for collaborative re-optimization, thereby generating an updated scheme more suitable for current and expected operating conditions. Finally, after rigorous update verification, the new parameters are safely deployed, and the next round of optimization iterations is initiated, forming a sustainable, self-improving intelligent closed loop.

[0064] Through the organic combination of these technical steps, this application effectively solves the key technical bottlenecks of traditional static planning methods, such as difficulty in adapting to dynamic operating environments, disconnect between capacity and strategy, and lack of long-term self-optimization capabilities. The final technical effect is systematic: it not only significantly improves the dynamic response accuracy and resource utilization efficiency of hybrid energy storage systems to the frequency regulation needs of the power grid at multiple time scales, but also effectively delays equipment wear and tear and improves the economic efficiency throughout the entire life cycle by optimizing operating conditions. Furthermore, it enables the entire system to have robustness and intelligence in dealing with complex operating conditions and continuous evolution, providing an advanced energy storage system solution to support the safe and stable operation of a high proportion of new energy power grids.

[0065] Example 2: The capacity optimization method for the multi-timescale hybrid energy storage system in the embodiments of this application has been described above. The capacity optimization system for the multi-timescale hybrid energy storage system in the embodiments of this application is described below. Please refer to [link / reference]. Figure 5 The capacity optimization system for the multi-timescale hybrid energy storage system in this application embodiment includes: The initial configuration module is used to establish and solve the initial capacity optimization model based on the multi-timescale decomposition results of new energy power fluctuations, and obtain the initial capacity configuration parameters and initial multi-timescale division of labor parameters of the hybrid energy storage system. The hybrid energy storage system includes a first type of energy storage unit and a second type of energy storage unit with complementary dynamic response characteristics. The frequency regulation operation and data acquisition module is used to control the hybrid energy storage system to participate in grid frequency regulation based on the initial capacity configuration parameters and the initial multi-time scale division of labor parameters, and to collect performance data and equipment status data during operation. The closed-loop feedback and reconfiguration module is used to generate performance feedback indicators based on performance data and equipment status data, dynamically correct the adjustable parameters of the capacity optimization model according to the performance feedback indicators, and re-solve the corrected capacity optimization model. In the re-solving process, the multi-timescale division of labor parameters and capacity configuration parameters are treated as correlation variables to obtain the updated capacity configuration parameters and updated multi-timescale division of labor parameters for the next running cycle. The application and iteration module is used to apply the updated capacity configuration parameters and updated multi-timescale division of labor parameters to the hybrid energy storage system, and return to the execution frequency regulation operation and data acquisition module to form a closed-loop optimization process.

[0066] Through the synergistic cooperation of the above-mentioned components, this application further improves the matching accuracy of capacity configuration and multi-timescale operation strategies of hybrid energy storage systems, the overall economic efficiency of operation, and the reliability and adaptability of long-term service.

[0067] This application also provides a capacity optimization device for a multi-timescale hybrid energy storage system. The capacity optimization device for the multi-timescale hybrid energy storage system includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor performs the steps of the capacity optimization method for the multi-timescale hybrid energy storage system in the above embodiments.

[0068] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0069] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0070] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A capacity optimization method for a multi-timescale hybrid energy storage system, characterized in that, The method includes: Step S1: Based on the multi-timescale decomposition results of new energy power fluctuations, establish and solve the initial capacity optimization model to obtain the initial capacity configuration parameters and initial multi-timescale division of labor parameters of the hybrid energy storage system. The hybrid energy storage system includes a first type of energy storage unit and a second type of energy storage unit with complementary dynamic response characteristics. Step S2: Based on the initial capacity configuration parameters and the initial multi-timescale division of labor parameters, control the hybrid energy storage system to participate in grid frequency regulation, and collect performance data and equipment status data during operation; Step S3: Based on the performance data and equipment status data, generate performance feedback indicators, and dynamically correct the adjustable parameters of the capacity optimization model according to the performance feedback indicators, and re-solve the corrected capacity optimization model. In the re-solve process, the multi-timescale division of labor parameters and the capacity configuration parameters are processed as correlation variables to obtain the updated capacity configuration parameters and updated multi-timescale division of labor parameters for the next running cycle. Step S4: Apply the updated capacity configuration parameters and the updated multi-timescale division of labor parameters to the hybrid energy storage system, and return to execute step S2 to form a closed-loop optimization process.

2. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 1, characterized in that, Step S1 further includes: The multi-timescale decomposition results include high-frequency components, mid-frequency components, and low-frequency components; the multi-timescale division of labor parameters are used to define the power allocation ratio of the first type of energy storage unit and the second type of energy storage unit to the power components of different timescales. The solution result of the capacity optimization model is configured as follows: the high-frequency component is mainly undertaken by the first type of energy storage unit, the low-frequency component is mainly undertaken by the second type of energy storage unit, and the medium-frequency component is jointly undertaken by the first type of energy storage unit and the second type of energy storage unit according to the multi-time scale division of labor parameters.

3. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 1, characterized in that, In step S3, dynamically correcting the adjustable parameters of the capacity optimization model includes adjusting at least one of the following parameters: The weighting coefficients of each sub-objective function related to frequency performance, investment cost, and equipment life loss in the capacity optimization model; Boundary conditions in the capacity optimization model regarding the operational safety constraints of the first type of energy storage unit and / or the second type of energy storage unit; The initial or reference values ​​of the multi-timescale division of labor parameters used in the re-solution process.

4. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 3, characterized in that, In step S3, the multi-timescale division of labor parameters and the capacity configuration parameters are processed as correlation variables, and this can be achieved through any of the following methods: Method 1: The capacity configuration parameters and the multi-timescale division of labor parameters are used together as decision variables and simultaneously optimized and determined during the re-solution process; Method 2: First, based on the performance feedback index, correct the multi-timescale division of labor parameters to obtain the corrected division of labor parameters. Then, using the corrected multi-timescale division of labor parameters as fixed conditions, solve the simplified model in the capacity optimization model that only contains capacity configuration variables to determine the updated capacity configuration parameters.

5. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 1, characterized in that, The execution of step S3 is controlled by any of the following conditions: The time control parameters reach the preset fixed cycle duration; If any of the performance feedback metrics exceeds the corresponding warning threshold; An update instruction was received from an external scheduling system.

6. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 1, characterized in that, The process includes an update verification step after step S3 and before step S4: Based on the updated capacity configuration parameters and the updated multi-timescale division of labor parameters, the hybrid energy storage system is subjected to a short-term trial operation. During the short-term trial operation, the operating performance indicators and equipment status indicators of the hybrid energy storage system are monitored and collected. If both the operating performance indicators and the equipment status indicators meet the preset admission criteria, then the updated capacity configuration parameters and the updated multi-timescale division of labor parameters will be confirmed to be applied to subsequent frequency regulation operations. Otherwise, the operation will be triggered to roll back to the capacity configuration parameters and multi-timescale division of labor parameters of the previous operating cycle, or the updated capacity configuration parameters and the updated multi-timescale division of labor parameters will be limited and then re-verified. The admission criteria are a set of pre-set quantitative or qualitative standards used to determine whether the capacity configuration parameters and multi-timescale division of labor parameters for the next operating cycle can be formally put into operation.

7. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 1, characterized in that, In step S3, before dynamically correcting the adjustable parameters of the capacity optimization model based on the performance feedback indicators, a performance root cause analysis step is also included: The performance data and equipment status data are fused and analyzed to generate a feature vector characterizing the overall status of the hybrid energy storage system. The feature vector is then matched and analyzed with a pre-set diagnostic rule base, which defines the mapping relationship between different abnormal state patterns and potential root causes. Based on the matching analysis, a diagnostic conclusion indicating the potential root cause of the current frequency regulation performance deviation is output; the types of the diagnostic conclusion include insufficient total frequency regulation capacity configuration of the hybrid energy storage system, unreasonable multi-timescale division of labor ratio between the first type of energy storage unit and the second type of energy storage unit, or excessive operating load of any type of energy storage unit; Based on the diagnostic conclusion, a targeted correction strategy corresponding to the type of the diagnostic conclusion is selected and executed to correct the adjustable parameters of the capacity optimization model.

8. The capacity optimization method for a multi-timescale hybrid energy storage system according to claim 1, characterized in that, In step S2, controlling the hybrid energy storage system to participate in grid frequency regulation adopts a hierarchical coordinated control strategy, including: The upper-level coordinator generates a total frequency regulation power command and power allocation commands for each energy storage unit based on the grid frequency deviation, frequency change rate, and real-time state margins of the first and second types of energy storage units. The lower-level actuator performs rapid tracking control on the power allocation commands. When the upper-level coordinator determines that the real-time state margin of any type of energy storage unit is lower than a safety threshold, it initiates a power command reallocation logic and reallocates the portion of the power commands that the energy storage unit with the real-time state margin lower than the safety threshold cannot safely handle to another type of energy storage unit or a backup resource.

9. A capacity optimization system for a multi-timescale hybrid energy storage system, used to implement the capacity optimization method for a multi-timescale hybrid energy storage system as described in any one of claims 1-8, characterized in that, The system includes: The initial configuration module is used to establish and solve the initial capacity optimization model based on the multi-timescale decomposition results of new energy power fluctuations, and obtain the initial capacity configuration parameters and initial multi-timescale division of labor parameters of the hybrid energy storage system. The hybrid energy storage system includes a first type of energy storage unit and a second type of energy storage unit with complementary dynamic response characteristics. The frequency regulation operation and data acquisition module is used to control the hybrid energy storage system to participate in grid frequency regulation based on the initial capacity configuration parameters and the initial multi-time scale division of labor parameters, and to collect performance data and equipment status data during operation. The closed-loop feedback and reconfiguration module is used to generate performance feedback indicators based on the performance data and equipment status data, dynamically correct the adjustable parameters of the capacity optimization model according to the performance feedback indicators, and re-solve the corrected capacity optimization model. In the re-solving process, the multi-timescale division of labor parameters and the capacity configuration parameters are processed as correlation variables to obtain the updated capacity configuration parameters and updated multi-timescale division of labor parameters for the next running cycle. The application and iteration module is used to apply the updated capacity configuration parameters and the updated multi-timescale division of labor parameters to the hybrid energy storage system, and return to execute the frequency regulation operation and data acquisition module to form a closed-loop optimization process.

10. A capacity optimization device for a multi-timescale hybrid energy storage system, characterized in that, The device includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the device to execute the capacity optimization method for a multi-timescale hybrid energy storage system as described in any one of claims 1-8.

Images