Data center flywheel energy storage system and method based on lcl filter

Abstract

The application relates to the technical field of data center energy storage, and discloses a data center flywheel energy storage system and method based on an LCL filter, which comprises the following steps: collecting a power grid side current signal, decomposing a harmonic component and dynamically adjusting LCL filter parameters to suppress the harmonic; determining a charge-discharge control strategy according to a direct-current bus voltage and a flywheel rotating speed and executing the control strategy; monitoring flywheel vibration and acceleration in real time to evaluate operation stability and correct the control strategy; and combining rotating speed feedback and voltage data compensation filter parameters to optimize the harmonic suppression effect. The system comprises a harmonic suppression module, a charge-discharge control module, a stability evaluation module and a filter parameter optimization module. Through multi-parameter dynamic adjustment and multi-module collaborative control, the application realizes composite suppression of power grid harmonics and accurate adjustment of the flywheel energy storage state, improves system operation stability and filter effect, and is suitable for a data center energy storage scene.

Description

Technical Field

[0001] This invention relates to the field of data center energy storage technology, specifically to a data center flywheel energy storage system and method based on an LCL filter. Background Technology

[0002] With the continuous expansion of data center scale and the explosive growth in computing power demand, extremely high requirements are placed on the stability, reliability, and power quality of power supply. Traditional data center energy storage solutions mostly use battery energy storage systems, but battery energy storage has problems such as limited lifespan, high maintenance costs, poor environmental adaptability, and potential safety hazards (such as fire risks), making it difficult to meet the data center's demand for long-cycle, high-reliability energy storage. Flywheel energy storage systems, with their advantages of high power density, fast response capability, long cycle life, and environmental friendliness, have become a research hotspot in the field of data center energy storage.

[0003] However, flywheel energy storage systems face numerous technical challenges in practical applications. On one hand, the current input from the grid often contains abundant harmonic components. These harmonics can cause a shift in the resonant characteristics of the LCL filter, affecting its filtering effect and consequently leading to grid pollution and reduced operating efficiency of the energy storage system. Traditional LCL filter parameter settings are typically based on fixed grid conditions and cannot adapt to dynamic changes in grid harmonic components, resulting in unstable filtering performance. On the other hand, the charging and discharging control of flywheel energy storage systems requires precise coordination of the relationship between the DC bus voltage, flywheel speed, and energy conversion modules. Existing control strategies mostly employ fixed control parameters and logic, making it difficult to dynamically adjust based on real-time operating conditions. This can easily lead to problems such as excessive power fluctuations and low energy storage efficiency during charging and discharging. Furthermore, during high-speed rotation of the flywheel rotor, changes in mechanical vibration and acceleration directly affect the system's operational stability and lifespan. Traditional stability assessment methods are usually based on threshold judgments, lacking real-time analysis and dynamic modeling of vibration data. This makes it impossible to provide early warnings of potential mechanical failures and to finely adjust the control strategy based on vibration characteristics.

[0004] In existing technologies, while some studies have proposed filtering methods based on single-parameter adjustment and fixed-mode charge-discharge control strategies for harmonic suppression and control strategy optimization in flywheel energy storage systems, these methods fail to fully consider the diversity and time-varying nature of grid harmonics, as well as the complex characteristics of multivariate coupling in flywheel energy storage systems. For example, traditional LCL filters can only suppress harmonics of specific frequencies and cannot simultaneously handle multiple types of harmonic components such as low-frequency resonance, fundamental components, and broadband oscillations. Existing charge-discharge control strategies lack in-depth analysis of the dynamic mapping relationship between DC bus voltage and flywheel speed, making it difficult to achieve precise control of charge-discharge power. In terms of system stability assessment, existing technologies have failed to effectively integrate mechanical vibration data with control strategies, resulting in a lack of scientific basis for modifying control strategies.

[0005] Therefore, designing a flywheel energy storage system and method that can dynamically adapt to changes in grid harmonics, achieve precise control of flywheel charging and discharging, and optimize control strategies based on real-time vibration data has become a key issue that urgently needs to be addressed in the field of data center energy storage technology. This invention aims to construct an efficient, stable, and reliable flywheel energy storage system by introducing multi-parameter dynamic adjustment of LCL filters, a charging and discharging control strategy based on real-time data, and a vibration-driven stability assessment and correction mechanism, in order to meet the high-quality energy storage requirements of data centers. Summary of the Invention

[0006] The purpose of this invention is to provide a data center flywheel energy storage system and method based on an LCL filter to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a data center flywheel energy storage method based on an LCL filter, the method comprising:

[0008] S1: The current signal input from the power grid side is acquired in real time through the current signal acquisition module to obtain the initial current dataset. The harmonic components in the initial current dataset are decomposed using the harmonic analysis algorithm to generate the fundamental component set and the harmonic component set. Based on the spectral characteristics of the fundamental component set and the harmonic component set, the topology parameters of the LCL filter are dynamically adjusted to obtain the first filter parameter set, the second filter parameter set and the third filter parameter set. The real-time current signal input from the power grid side is then subjected to harmonic suppression processing based on the above parameter sets.

[0009] S2: Obtain the DC bus voltage data of the flywheel energy storage device through the voltage detection module, combine it with the real-time speed information of the flywheel rotor, conduct a feasibility analysis on the flywheel charging and discharging state corresponding to each filter parameter group, determine the target control strategy of the flywheel energy storage system based on the analysis results, and execute the target control strategy through the energy conversion module to adjust the energy storage state of the flywheel rotor.

[0010] S3: When executing the target control strategy, the flywheel state monitoring module acquires the acceleration and mechanical vibration data of the flywheel rotor in real time, dynamically evaluates the operational stability of the flywheel energy storage system, determines whether the target control strategy needs to be modified based on the evaluation results, and optimizes the parameters of the modified control strategy.

[0011] S4: The energy conversion module updates the energy storage state of the flywheel rotor according to the modified control strategy, obtains the updated flywheel speed data through the flywheel speed feedback module, and performs real-time compensation on the dynamic response characteristics of the LCL filter in combination with the DC bus voltage data. The compensated filter parameters are then transmitted to the grid-side control terminal to optimize the harmonic suppression effect.

[0012] Preferably, S1 includes:

[0013] S11: The current signal input from the grid side is sampled in segments. The sampled data is compared with the preset fundamental reference waveform. The sampling segments with deviations exceeding the threshold are marked as harmonic interference segments. A harmonic component set is generated. The high-frequency components in the harmonic component set are feature extracted according to the standard resonant frequency of the LCL filter.

[0014] S12: Select harmonic components with low-frequency resonance characteristics from the harmonic component set, and use an adaptive weighting algorithm to iteratively optimize the inductance and capacitance parameters of the LCL filter, and use the optimized inductance parameters as the first filter parameter group.

[0015] The damping resistance parameters of the LCL filter are dynamically matched based on the amplitude change rate of the fundamental component set, and the matched resistance parameters are used as the second set of filter parameters.

[0016] Harmonic components with broadband oscillation characteristics are extracted from the harmonic component set. The parallel capacitor parameters of the LCL filter are corrected in real time by the phase compensation algorithm, and the corrected capacitor parameters are used as the third filter parameter group.

[0017] S13: Perform spectrum segmentation on the current signal input from the grid side at time t, and input the segmented low-frequency resonant component, fundamental component and wideband oscillation component into the filter channels corresponding to the first filter parameter group, the second filter parameter group and the third filter parameter group respectively. Perform composite harmonic suppression on the grid side current through multi-channel parallel filtering.

[0018] Preferably, S2 includes:

[0019] S21: Install multiple sets of voltage sensors at the DC bus end of the flywheel energy storage device, collect voltage fluctuation data of the DC bus through the voltage sensors, and construct a dynamic energy model of the flywheel energy storage system based on the fluctuation data. In the model, the rated speed of the flywheel rotor is used as the reference value to establish a speed-voltage mapping relationship.

[0020] S22: Obtain the speed change rate of the flywheel rotor in the previous control cycle, and deduce the charging and discharging power of the flywheel energy storage system in reverse by combining the DC bus voltage data. If no charging or discharging command is detected in the current control cycle, the target speed of the flywheel rotor is set to the maintenance value. If a charging or discharging command is detected, the target speed is adjusted stepwise according to the command type.

[0021] S23: Based on the deviation between the target speed and the actual speed, the torque compensation amount of the flywheel drive motor is calculated by the proportional-integral algorithm. The feasibility of the compensation amount is verified by combining the dynamic energy model. If the verification is successful, the compensation amount is used as the core parameter of the target control strategy. Otherwise, the speed adjustment gradient is regenerated and the target control strategy is updated.

[0022] S24: The energy conversion module drives the flywheel rotor to accelerate or decelerate according to the target control strategy in order to achieve precise adjustment of the energy storage state.

[0023] Preferably, S3 includes:

[0024] S31: The axial and radial vibration spectra of the flywheel rotor are collected by vibration sensors. The imbalance of the flywheel mechanical structure is quantitatively analyzed based on the spectrum energy distribution, and vibration evaluation coefficients are generated.

[0025] S32: Based on the correlation between the vibration evaluation coefficient and the flywheel speed, a dynamic stability criterion model is constructed. If the model output value exceeds the preset threshold, it is determined that the torque compensation in the target control strategy needs to be attenuated and corrected.

[0026] S33: The flywheel status monitoring module tracks the corrected control strategy in real time. If the vibration evaluation coefficient does not exceed the standard within three consecutive sampling cycles after correction, the corrected strategy is determined to be effective; otherwise, the flywheel emergency braking protection mechanism is triggered.

[0027] Preferably, S33 includes:

[0028] The current limit of the flywheel drive motor is dynamically adjusted based on the corrected torque compensation. If the adjusted current value exceeds the rated capacity of the motor, the backup energy storage unit is activated to assist in the control of the flywheel speed until the mechanical vibration of the flywheel returns to a safe range.

[0029] Preferably, the present invention also includes a system for the above-mentioned data center flywheel energy storage method based on LCL filter, the system including a harmonic suppression module, a flywheel charging and discharging control module, a dynamic stability evaluation module and a filter parameter optimization module;

[0030] The harmonic suppression module is used to generate multiple sets of filtering parameters through harmonic analysis algorithms and to perform composite harmonic suppression on the grid-side current.

[0031] The flywheel charging and discharging control module is used to determine the target control strategy based on the DC bus voltage data and flywheel speed information, and drive the energy conversion module to perform charging and discharging operations.

[0032] The dynamic stability assessment module is used to monitor flywheel mechanical vibration data in real time and dynamically correct the control strategy.

[0033] The filter parameter optimization module is used to compensate the dynamic response of the LCL filter in real time by combining the flywheel speed feedback data.

[0034] Preferably, the harmonic suppression module includes a harmonic feature extraction unit, a parameter dynamic adjustment unit, and a multi-channel filtering unit;

[0035] The harmonic feature extraction unit performs spectrum segmentation and feature labeling on the grid-side current signal;

[0036] The parameter dynamic adjustment unit iteratively optimizes the inductance, capacitance, and resistance parameters of the LCL filter based on harmonic characteristics.

[0037] The multi-channel filtering unit performs graded suppression of different types of harmonic components through parallel filtering channels.

[0038] Preferably, the flywheel charging and discharging control module includes a speed-voltage mapping unit, a power reverse derivation unit, and a torque compensation calculation unit;

[0039] The speed-voltage mapping unit constructs a dynamic energy model based on DC bus voltage fluctuation data;

[0040] The power reverse derivation unit calculates the charging and discharging power requirements based on the flywheel speed change rate.

[0041] The torque compensation calculation unit generates the torque adjustment amount of the drive motor through a proportional-integral algorithm.

[0042] Preferably, the dynamic stability assessment module includes a vibration spectrum analysis unit, a stability criterion construction unit, and a strategy correction unit;

[0043] The vibration spectrum analysis unit quantifies and evaluates the mechanical vibration energy of the flywheel rotor;

[0044] The stability criterion construction unit generates a dynamic stability threshold based on vibration data;

[0045] The strategy correction unit performs parameter decay correction on control strategies that exceed the threshold.

[0046] Preferably, the filter parameter optimization module uses the coupling relationship between flywheel speed feedback data and DC bus voltage to perform real-time compensation on the damping resistance and parallel capacitor parameters of the LCL filter, so as to reduce the harmonic distortion rate of the grid-side current.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] In terms of harmonic suppression, the current signal acquisition module collects the grid-side current signal in real time, and uses a harmonic analysis algorithm to decompose the fundamental component set and harmonic component sets. Based on their spectral characteristics, the topology parameters of the LCL filter are dynamically adjusted to generate multiple sets of filter parameters. By using multi-channel parallel filtering, harmonic components with different characteristics (such as low-frequency resonant components and broadband oscillation components) are targeted for suppression. Compared with traditional single-parameter filters, this method can more effectively cope with complex harmonic environments, significantly reduce the harmonic distortion rate of the grid-side current, improve power quality, and reduce the damage of harmonics to data center equipment.

[0049] In the flywheel charging and discharging control process, a dynamic energy model is constructed by combining DC bus voltage data and real-time flywheel rotor speed information, and a speed-voltage mapping relationship is established. This allows for precise analysis of the feasibility of the flywheel charging and discharging states corresponding to each set of filter parameters, thereby determining a reasonable target control strategy. The torque compensation amount is calculated using a proportional-integral algorithm, and its feasibility is verified using the dynamic energy model, achieving precise adjustment of the flywheel rotor's energy storage state. This control method fully considers the actual operating state of the system, improves the accuracy and response speed of charging and discharging power control, and enhances the energy conversion efficiency of the flywheel energy storage system.

[0050] In terms of system stability assessment and control strategy correction, the flywheel condition monitoring module acquires real-time acceleration and mechanical vibration data of the flywheel rotor. The vibration spectrum is analyzed, and the degree of mechanical structural imbalance is quantitatively assessed to construct a dynamic stability criterion model. When system stability anomalies occur, the target control strategy can be promptly corrected, such as attenuating torque compensation. If necessary, a backup energy storage unit can be activated for auxiliary control to ensure the flywheel mechanical vibration returns to a safe range. This dynamic assessment and correction mechanism effectively improves the system's operational stability and reliability, and reduces safety risks caused by abnormal mechanical vibration.

[0051] Regarding filter parameter optimization, the damping resistor and parallel capacitor parameters of the LCL filter are compensated in real time by combining the coupling relationship between flywheel speed feedback data and DC bus voltage, thereby optimizing the filter's dynamic response characteristics. This real-time compensation mechanism enables the filter to dynamically adjust its parameters according to the operating status of the flywheel energy storage system, further improving the harmonic suppression effect, forming a positive interaction between the flywheel energy storage system and the filter system, and enhancing the overall system's collaborative performance.

[0052] Furthermore, this invention employs a modular design, dividing the system into multiple functional modules, including a harmonic suppression module, a flywheel charging and discharging control module, a dynamic stability assessment module, and a filter parameter optimization module. These modules have clearly defined roles and work collaboratively, resulting in excellent scalability and maintainability for the entire system. Simultaneously, the advanced algorithms and technologies used in each module (such as adaptive weighting algorithms, phase compensation algorithms, and proportional-integral algorithms) enhance the system's intelligence and automation levels, reduce manual intervention, lower operating costs, and provide strong support for the stable and efficient operation of the data center. Attached Figure Description

[0053] Figure 1 This is a schematic diagram illustrating the working principle of the data center flywheel energy storage system and method based on LCL filters described in this invention.

[0054] Figure 2 A schematic diagram illustrating the working principle of adaptive optimization of LCL filter parameters for multi-band harmonic separation;

[0055] Figure 3 A flowchart illustrating the charging and discharging control strategy for a flywheel energy storage system based on DC bus voltage fluctuations;

[0056] Figure 4 This is a flowchart of an emergency control system for a flywheel energy storage system based on current limiting adjustment.

[0057] Figure 5 This is a schematic diagram illustrating the working principle of the harmonic suppression module. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Please see Figures 1-5 This invention provides a data center flywheel energy storage method based on an LCL filter, specifically including the following steps:

[0060] S1: The current signal input from the power grid side is acquired in real time through the current signal acquisition module to obtain an initial current dataset. The harmonic components in the initial current dataset are decomposed using a harmonic analysis algorithm to generate a fundamental component set and a harmonic component set. Based on the spectral characteristics of the fundamental and harmonic component sets, the topology parameters of the LCL filter are dynamically adjusted to obtain a first set of filtering parameters, a second set of filtering parameters, and a third set of filtering parameters. Harmonic suppression processing is then performed on the real-time current signal input from the power grid side based on these parameter sets.

[0061] S2: Obtain the DC bus voltage data of the flywheel energy storage device through the voltage detection module, combine it with the real-time speed information of the flywheel rotor, perform a feasibility analysis on the flywheel charging and discharging state corresponding to each filter parameter group, determine the target control strategy of the flywheel energy storage system based on the analysis results, and execute the target control strategy through the energy conversion module to adjust the energy storage state of the flywheel rotor.

[0062] S3: When executing the target control strategy, the flywheel state monitoring module acquires the acceleration and mechanical vibration data of the flywheel rotor in real time, dynamically evaluates the operational stability of the flywheel energy storage system, determines whether the target control strategy needs to be modified based on the evaluation results, and optimizes the parameters of the modified control strategy.

[0063] S4: The energy conversion module updates the energy storage state of the flywheel rotor according to the modified control strategy, obtains the updated flywheel speed data through the flywheel speed feedback module, and performs real-time compensation on the dynamic response characteristics of the LCL filter in combination with the DC bus voltage data. The compensated filter parameters are then transmitted to the grid-side control terminal to optimize the harmonic suppression effect.

[0064] The present invention will be further described below with reference to Examples 1 to 5:

[0065] Example 1:

[0066] Based on the overall implementation scheme described above, this embodiment further defines S1. The current signal acquisition module uses a high-precision current transformer to sample the current signal input from the power grid side in segments. The sampling period is set to a fixed duration according to the standard frequency of the power grid and the range of common harmonic frequencies to ensure coverage of the periodic characteristics of harmonic signals of different frequency components. The discrete current data obtained after sampling constitutes the initial current dataset. The system compares the sampled data with the preset fundamental reference waveform point by point. Through preset deviation calculation rules, the sampling segments with absolute deviation values ​​exceeding a preset threshold are automatically marked as harmonic interference segments. Unmarked sampling segments are included in the fundamental component set. The data of the marked harmonic interference segments are extracted separately to generate the harmonic component set.

[0067] For the harmonic component set, the system first uses the standard resonant frequency of the LCL filter as the frequency screening threshold. A bandpass filtering algorithm from digital signal processing is then used to extract features from the high-frequency components in the harmonic component set, separating high-frequency harmonic feature data above the standard resonant frequency. Next, harmonic components with frequencies below the standard resonant frequency and exhibiting low-frequency resonance characteristics are further screened from the harmonic component set. Specifically, low-frequency resonance characteristics are determined by analyzing the spectral peak distribution and phase consistency of the harmonic components. For these harmonic components with low-frequency resonance characteristics, an adaptive weighting algorithm is used to iteratively optimize the inductor and capacitor parameters of the LCL filter. The algorithm aims to minimize the spectral energy of the low-frequency resonance characteristics. By setting initial inductor and capacitor parameter values, the weighting factor is dynamically adjusted based on the current harmonic component's feature parameters, iteratively updating the inductor and capacitor parameters until the spectral energy of the low-frequency resonance characteristics converges within a preset range. The final optimized inductor parameters are then used as the first set of filter parameters.

[0068] Based on the amplitude change rate of the fundamental component set, the system dynamically matches the damping resistor parameters of the LCL filter. Specifically, the amplitude change rate is obtained by calculating the ratio of the difference in amplitude change of the fundamental component in adjacent sampling periods to the time interval. According to the magnitude and direction of the amplitude change rate, the corresponding damping resistor parameter value is found through a preset damping resistor matching rule table, and the matched resistor parameters are used as the second set of filter parameters. For harmonic components with broadband oscillation characteristics in the harmonic component set, the system performs real-time correction of the parallel capacitor parameters of the LCL filter through a phase compensation algorithm: First, the phase offset of the broadband oscillation characteristic harmonic component is analyzed. According to the magnitude and direction of the phase offset, the correction value of the parallel capacitor parameter is calculated through a pre-established phase-capacitance correction mapping relationship. The current parallel capacitor parameters are adjusted in real time, and the corrected capacitor parameters are used as the third set of filter parameters.

[0069] When processing the current signal input from the grid side at time t, the system first performs spectral segmentation on the current signal using a Fast Fourier Transform (FFT) algorithm, decomposing the spectrum into components of different frequency ranges. Based on the harmonic characteristic frequency ranges corresponding to each filter parameter group, the segmented low-frequency resonant component, fundamental component, and wideband oscillation component are routed to the filter channels corresponding to the first, second, and third filter parameter groups, respectively. The three filter channels are connected in parallel, with each channel independently filtering the current component in its corresponding frequency range: the first filter channel uses optimized inductance parameters to suppress the low-frequency resonant component; the second filter channel uses matched damping resistor parameters to dampen the fluctuations of the fundamental component; and the third filter channel uses corrected parallel capacitor parameters to perform phase compensation and amplitude attenuation of the wideband oscillation component. Through multi-channel parallel filtering, the system achieves composite suppression of different types of harmonic components in the grid-side current. The filtering results from each channel are combined at the output and returned to the grid side, completing the harmonic suppression process. Throughout the process, the system monitors the parameter status and current signal characteristics of each filter channel in real time, and dynamically adjusts the parameter optimization and spectrum segmentation strategies based on real-time data to ensure the effectiveness and adaptability of harmonic suppression.

[0070] Example 2:

[0071] Multiple sets of voltage sensors are spaced along the current flow direction at the DC bus end of the flywheel energy storage device. Each set of sensors employs a redundant design to ensure the reliability of the acquired data. The system preprocesses the acquired voltage data, using a moving average filtering algorithm to eliminate random noise interference, resulting in smooth DC bus voltage fluctuation data. Based on this fluctuation data, a dynamic energy model of the flywheel energy storage system is constructed. The model uses the rated speed of the flywheel rotor as the baseline value to establish a speed-voltage mapping relationship. Specifically, a polynomial fitting algorithm is used to fit the speed in historical operating data with the corresponding DC bus voltage value to obtain a continuous mapping function relationship.

[0072] The system acquires the rate of change of the flywheel rotor speed during the previous control cycle and, combined with the current DC bus voltage data, uses the principle of energy conservation to inversely deduce the charging and discharging power of the flywheel energy storage system. Specifically, the system calculates the energy storage state corresponding to the current speed based on the dynamic energy model, compares it with the energy storage state of the previous cycle, and calculates the energy change based on the time interval, thus obtaining the charging and discharging power. If no external charging or discharging command is detected during the current control cycle, the system automatically sets the target speed of the flywheel rotor as a maintenance value. This maintenance value is dynamically adjusted based on the flywheel system's self-discharge rate and the energy required to maintain basic system operation. If a charging or discharging command is detected, the system makes a step adjustment to the target speed based on the command type (charging or discharging) and command strength. The step amplitude is calculated based on the command power requirement and the current system state.

[0073] Based on the deviation between the target speed and the actual speed, the system calculates the torque compensation amount for the flywheel drive motor using a proportional-integral (PI) algorithm. The algorithm proportionally amplifies and integrally accumulates the deviation to generate a corresponding control signal as the torque compensation amount. To ensure the feasibility of the compensation amount, the system verifies it using a dynamic energy model. The specific verification process is as follows: The energy change trend of the system after applying the torque compensation amount is calculated based on the dynamic energy model, and it is assessed whether it exceeds the system's physical limitations and safety boundaries. If the verification passes, the compensation amount is used as the core parameter of the target control strategy; if the verification fails, the system regenerates the speed adjustment gradient, reduces the step amplitude, and recalculates and verifies the torque compensation amount until a feasible control parameter is found.

[0074] After receiving the target control strategy, the energy conversion module controls the flywheel drive motor via a power electronic converter. For charging commands, the system increases the motor's input voltage, causing it to output positive torque and drive the flywheel rotor to accelerate, converting electrical energy into stored mechanical energy. For discharging commands, the system decreases the motor's input voltage, causing it to output reverse torque, causing the flywheel rotor to decelerate and release mechanical energy, which is then converted back into electrical energy for output. Throughout the process, the system monitors the flywheel rotor's speed and DC bus voltage in real time, using closed-loop feedback control to ensure the precise execution of the target control strategy and achieve accurate adjustment of the flywheel energy storage system's charging and discharging state. Simultaneously, the system continuously updates the dynamic energy model parameters to adapt to performance drift caused by flywheel system aging, environmental temperature changes, and other factors, ensuring the adaptability and stability of the control strategy.

[0075] Example 3:

[0076] Axial and radial vibration signals of the flywheel rotor during operation are collected using vibration sensors. The sensors employ a triaxial accelerometer, sampling the vibration signals in real time at a fixed sampling frequency to obtain time-domain vibration data. A Fast Fourier Transform (FFT) is performed on the time-domain vibration data to convert it into a frequency-domain vibration spectrum. Based on the energy distribution of each frequency component in the spectrum, the proportion of vibration energy in different frequency intervals is calculated, thereby quantifying the degree of imbalance in the flywheel mechanical structure. Specifically, vibration assessment coefficients are generated using the following formula. :

[0077]

[0078] in, For the first The vibration energy amplitude within a characteristic frequency range, which is preset according to the mechanical characteristics of the flywheel rotor; Aj is the vibration energy amplitude within the j-th characteristic frequency range; The weighting coefficients are for the corresponding characteristic frequency ranges. The weighting coefficients are determined based on the degree of influence of vibration on the stability of the flywheel system within that frequency range. For example, the frequency range corresponding to the critical speed of the flywheel rotor has a higher weighting coefficient. This represents the total number of characteristic frequency intervals.

[0079] According to the vibration assessment coefficient With flywheel speed Historical data was used to construct a dynamic stability criterion model. The model was fitted using a nonlinear regression algorithm. and The correlation was used to obtain the dynamic stability threshold that varies with rotational speed. When the model outputs the current vibration assessment coefficients... Exceeding the threshold at the corresponding speed If the flywheel energy storage system is deemed to have insufficient operational stability, the torque compensation in the target control strategy needs to be attenuated and corrected. The specific method of attenuation correction is as follows: based on the magnitude of exceeding the threshold, the absolute value of the torque compensation is reduced by a preset proportional coefficient, while keeping the torque direction unchanged, in order to reduce the acceleration of the flywheel rotor and reduce mechanical vibration energy.

[0080] The flywheel condition monitoring module tracks the modified control strategy in real time and continuously collects vibration evaluation coefficients for the next three sampling cycles. , , If the vibration evaluation coefficients in all three sampling periods do not exceed the dynamic stability threshold at the corresponding rotational speed... If the vibration assessment coefficient exceeds the standard in any sampling period, the correction strategy is deemed effective, and the system continues to operate with the current control parameters. If the vibration assessment coefficient still exceeds the standard in any sampling period, the correction strategy is deemed ineffective in improving system stability, triggering the flywheel emergency braking protection mechanism. The emergency braking protection mechanism cuts off the power to the flywheel drive motor and activates the mechanical braking device, causing the flywheel rotor to stop rotating within a short time, preventing mechanical damage or safety accidents caused by continuous vibration. Throughout the process, the system records vibration data, speed data, and control parameter adjustment processes in real time, providing data support for subsequent system maintenance and control strategy optimization.

[0081] Example 4:

[0082] When the system dynamically adjusts the current limit of the flywheel drive motor based on the corrected torque compensation, assuming the calculated current limit is 800A at a certain moment, while the motor's rated current capacity is 750A, the system determines that the adjusted current value exceeds the motor's rated capacity and immediately activates the backup energy storage unit to assist in controlling the flywheel speed. The backup energy storage unit uses a high-energy-density lithium battery pack, connected to the DC bus of the flywheel energy storage system via a DC / DC converter, and possesses the ability for rapid response and bidirectional energy flow.

[0083] The system first sends a start command to the backup energy storage unit via the communication interface to activate the battery management system (BMS). The BMS quickly detects the battery pack's status parameters, including SOC (State of Charge), individual cell voltage, and temperature. After confirming that the battery pack is in a dischargeable state, it adjusts the DC / DC converter's operating mode to boost mode, raising the battery pack voltage to a level matching the DC bus voltage. Simultaneously, the system reduces the main power input to the flywheel drive motor, decreasing the motor's current demand on the grid and transferring some energy demand to the backup energy storage unit.

[0084] During auxiliary control, the backup energy storage unit adjusts its output power in real time based on the flywheel speed deviation. For example, when the flywheel speed is lower than the target value, the backup energy storage unit increases its output power, providing additional electrical energy to the flywheel drive motor through the power electronic converter, helping the motor generate greater torque and accelerating the flywheel rotor. Conversely, when the flywheel speed is higher than the target value, the backup energy storage unit reduces its output power or switches to charging mode to absorb the energy released by the flywheel, thus achieving fine-tuning of the flywheel speed.

[0085] The system assesses the effectiveness of the auxiliary control by monitoring the vibration data of the flywheel rotor in real time. For example, vibration signals are collected by an accelerometer mounted on the flywheel bearing housing, and the vibration spectrum is obtained after signal processing. If the spectrum analysis shows that the axial vibration frequency of the flywheel rotor gradually decreases from the original 35Hz to 25Hz, and the vibration amplitude decreases, it indicates that the auxiliary control strategy is effective and the imbalance of the flywheel mechanical structure is being reduced.

[0086] As the flywheel's mechanical vibration returns to a safe range, the system gradually reduces the involvement of the backup energy storage unit. For example, when the characteristic frequency and amplitude in the vibration spectrum remain within the safe threshold for three consecutive sampling periods, the system begins to reduce the output power of the backup energy storage unit while gradually increasing the main power input power of the flywheel drive motor, allowing the flywheel speed control to gradually return to normal. During this process, the system uses a smooth transition algorithm to ensure the stability of the flywheel speed during power switching, avoiding speed fluctuations and vibration rebounds caused by sudden power changes.

[0087] When the output power of the backup energy storage unit drops to zero, the system completely shuts down the discharge channel of the backup energy storage unit, and the BMS switches to standby mode to maintain real-time monitoring of the battery pack status. Throughout the process, the system continuously tracks the vibration data and speed information of the flywheel rotor to ensure that the flywheel energy storage system can still operate stably near the target operating point after the backup energy storage unit exits auxiliary control.

[0088] Example 5:

[0089] The system comprises a harmonic suppression module, a flywheel charging and discharging control module, a dynamic stability evaluation module, and a filter parameter optimization module. These modules interact and collaborate in real-time via a data bus. The harmonic suppression module consists of a harmonic feature extraction unit, a parameter dynamic adjustment unit, and a multi-channel filtering unit. When performing spectral segmentation on the grid-side current signal, the harmonic feature extraction unit employs a wavelet transform-based multi-resolution analysis method to decompose the current signal into different frequency bands. By setting different wavelet basis functions and decomposition levels, it achieves accurate extraction of harmonic features across different frequency ranges. When optimizing the inductance parameters of the LCL filter, the parameter dynamic adjustment unit uses a particle swarm optimization algorithm. With harmonic suppression effect as the objective function and inductance value as the optimization variable, it iteratively optimizes the parameters by simulating bird foraging behavior. In each iteration, the particle position is updated based on individual and global extrema, gradually approaching the optimal inductance parameter value.

[0090] When constructing the dynamic energy model, the speed-voltage mapping unit of the flywheel charging and discharging control module considers the rotational inertia variation characteristics of the flywheel rotor. Specifically, it establishes a nonlinear mapping relationship including a rotational inertia correction coefficient by collecting voltage data at different speeds. This correction coefficient is dynamically adjusted according to the mechanical properties and temperature characteristics of the flywheel material. The power inverse derivation unit introduces a time-series prediction algorithm when calculating charging and discharging power requirements. Based on historical power data and the current system state, it predicts the power change trend in the near future and adjusts the control strategy in advance to cope with sudden load changes. The torque compensation calculation unit uses a fuzzy PID control algorithm. Based on the speed deviation and its rate of change, it dynamically adjusts the parameters of the PID controller through a fuzzy inference system to improve the system's response speed and anti-interference capability.

[0091] The vibration spectrum analysis unit of the dynamic stability assessment module preprocesses the acquired vibration signals by using Empirical Mode Decomposition (EMD) to decompose the signals into multiple Intrinsic Mode Functions (IMFs). Then, it performs spectral analysis on each IMF component to extract characteristic frequency components related to mechanical faults. The stability criterion construction unit, based on vibration spectrum characteristics, constructs a multi-dimensional stability assessment index system, including vibration energy distribution entropy and frequency component complexity. The weights of each index are determined using the Analytic Hierarchy Process (AHP) to form a comprehensive stability criterion. The strategy correction unit, when correcting the parameter attenuation of the control strategy, employs an adaptive learning mechanism to dynamically adjust the attenuation coefficient based on historical correction effects, ensuring that vibration is suppressed without affecting the system's dynamic response performance.

[0092] The filter parameter optimization module, based on the coupling relationship between flywheel speed feedback data and DC bus voltage, designs an adaptive Kalman filter to compensate for the LCL filter parameters in real time. This filter treats flywheel speed and DC bus voltage as state variables, and by establishing a state-space model, it adaptively adjusts the filter gain using the prediction error covariance matrix. Even with measurement noise and system uncertainties, it can accurately estimate the optimal filter parameters. The system also incorporates a parameter self-learning mechanism. By collecting long-term operating data, it uses a neural network algorithm to uncover the implicit relationships between flywheel speed, DC bus voltage, and filter parameters, further improving the accuracy and adaptability of parameter compensation.

[0093] Data interaction between modules employs a combination of time-triggered and event-triggered mechanisms. For example, after completing a harmonic feature extraction, the harmonic suppression module sends relevant data to the filter parameter optimization module via event triggering; while the flywheel charging and discharging control module transmits flywheel status data to the dynamic stability assessment module at fixed intervals via time triggering. The data bus adopts a redundant design, with two communication links, a primary and a backup. When the primary link fails, it automatically switches to the backup link to ensure the reliability of system communication. Simultaneously, the system uses distributed clock synchronization technology to ensure the time consistency of data acquisition and processing across modules, improving the overall system's collaborative control capabilities.

[0094] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0095] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A flywheel energy storage method for data centers based on an LCL filter, characterized in that: The method includes: S1: The current signal input from the power grid side is acquired in real time through the current signal acquisition module to obtain the initial current dataset. The harmonic components in the initial current dataset are decomposed using the harmonic analysis algorithm to generate the fundamental component set and the harmonic component set. Based on the spectral characteristics of the fundamental component set and the harmonic component set, the topology parameters of the LCL filter are dynamically adjusted to obtain the first filter parameter set, the second filter parameter set and the third filter parameter set. The real-time current signal input from the power grid side is then subjected to harmonic suppression processing based on the above parameter sets. S2: Obtain the DC bus voltage data of the flywheel energy storage device through the voltage detection module, combine it with the real-time speed information of the flywheel rotor, construct the dynamic energy model of the flywheel energy storage system based on the DC bus voltage fluctuation data, and execute the target control strategy through the energy conversion module to adjust the energy storage state of the flywheel rotor. S3: When executing the target control strategy, the mechanical vibration data of the flywheel rotor is acquired in real time through the flywheel state monitoring module to dynamically evaluate the operational stability of the flywheel energy storage system. Based on the evaluation results, it is determined whether the target control strategy needs to be modified, and the parameters of the modified control strategy are optimized. S4: The energy conversion module updates the energy storage state of the flywheel rotor according to the modified control strategy, obtains the updated flywheel speed data through the flywheel speed feedback module, and performs real-time compensation on the dynamic response characteristics of the LCL filter in combination with the DC bus voltage data. The compensated filter parameters are then transmitted to the grid-side control terminal to optimize the harmonic suppression effect. S1 includes: S11: The current signal input from the grid side is sampled in segments. The sampled data is compared with the preset fundamental reference waveform. The sampling segments with deviations exceeding the threshold are marked as harmonic interference segments. A harmonic component set is generated. The high-frequency components in the harmonic component set are feature extracted according to the standard resonant frequency of the LCL filter. S12: Select harmonic components with low-frequency resonance characteristics from the harmonic component set, and use an adaptive weighting algorithm to iteratively optimize the inductance and capacitance parameters of the LCL filter, and use the optimized inductance parameters as the first filter parameter group. The damping resistance parameters of the LCL filter are dynamically matched based on the amplitude change rate of the fundamental component set, and the matched resistance parameters are used as the second set of filter parameters. Harmonic components with broadband oscillation characteristics are extracted from the harmonic component set. The parallel capacitor parameters of the LCL filter are corrected in real time by the phase compensation algorithm, and the corrected capacitor parameters are used as the third filter parameter group. S13: Perform spectrum segmentation on the current signal input from the grid side at time t, and input the segmented low-frequency resonant component, fundamental component and wideband oscillation component into the filter channels corresponding to the first filter parameter group, the second filter parameter group and the third filter parameter group respectively. Perform composite harmonic suppression on the grid side current through multi-channel parallel filtering. S2 includes: S21: Install multiple sets of voltage sensors at the DC bus end of the flywheel energy storage device, collect voltage fluctuation data of the DC bus through the voltage sensors, and construct a dynamic energy model of the flywheel energy storage system based on the fluctuation data. In the model, the rated speed of the flywheel rotor is used as the reference value to establish a speed-voltage mapping relationship. S22: Obtain the speed change rate of the flywheel rotor in the previous control cycle, and deduce the charging and discharging power of the flywheel energy storage system in reverse by combining the DC bus voltage data. If no charging or discharging command is detected in the current control cycle, the target speed of the flywheel rotor is set to the maintenance value. If a charging or discharging command is detected, the target speed is adjusted stepwise according to the command type. S23: Based on the deviation between the target speed and the actual speed, the torque compensation of the flywheel drive motor is calculated using the proportional-integral algorithm, and the compensation is used as the core parameter of the target control strategy. S24: The energy conversion module drives the flywheel rotor to accelerate or decelerate according to the target control strategy in order to achieve precise adjustment of the energy storage state.

2. The data center flywheel energy storage method based on an LCL filter according to claim 1, characterized in that: S3 includes: S31: The axial and radial vibration spectra of the flywheel rotor are collected by vibration sensors. The imbalance of the flywheel mechanical structure is quantitatively analyzed based on the spectrum energy distribution, and vibration evaluation coefficients are generated. S32: Based on the correlation between the vibration evaluation coefficient and the flywheel speed, a dynamic stability criterion model is constructed. If the model output value exceeds the preset threshold, it is determined that the torque compensation in the target control strategy needs to be attenuated and corrected. S33: The flywheel status monitoring module tracks the corrected control strategy in real time. If the vibration evaluation coefficient does not exceed the standard within three consecutive sampling cycles after correction, the corrected strategy is determined to be effective; otherwise, the flywheel emergency braking protection mechanism is triggered.

3. The data center flywheel energy storage method based on an LCL filter according to claim 2, characterized in that: S33 includes: The current limit of the flywheel drive motor is dynamically adjusted based on the corrected torque compensation. If the adjusted current value exceeds the rated capacity of the motor, the backup energy storage unit is activated to assist in the control of the flywheel speed until the mechanical vibration of the flywheel returns to a safe range.

4. A system applied to the data center flywheel energy storage method based on an LCL filter as described in any one of claims 1-3, characterized in that: The system includes a harmonic suppression module, a flywheel charging and discharging control module, a dynamic stability evaluation module, and a filter parameter optimization module; The harmonic suppression module is used to generate multiple sets of filtering parameters through harmonic analysis algorithms and to perform composite harmonic suppression on the grid-side current. The flywheel charging and discharging control module is used to determine the target control strategy based on the DC bus voltage data and flywheel speed information, and drive the energy conversion module to perform charging and discharging operations. The dynamic stability assessment module is used to monitor flywheel mechanical vibration data in real time and dynamically correct the control strategy. The filter parameter optimization module is used to compensate the dynamic response of the LCL filter in real time by combining the flywheel speed feedback data.

5. A data center flywheel energy storage system based on an LCL filter according to claim 4, characterized in that: The harmonic suppression module includes a harmonic feature extraction unit, a parameter dynamic adjustment unit, and a multi-channel filtering unit; The harmonic feature extraction unit performs spectrum segmentation and feature labeling on the grid-side current signal; The parameter dynamic adjustment unit iteratively optimizes the inductance, capacitance, and resistance parameters of the LCL filter based on harmonic characteristics. The multi-channel filtering unit performs graded suppression of different types of harmonic components through parallel filtering channels.

6. A data center flywheel energy storage system based on an LCL filter according to claim 5, characterized in that: The flywheel charging and discharging control module includes a speed-voltage mapping unit, a power reverse derivation unit, and a torque compensation calculation unit. The speed-voltage mapping unit constructs a dynamic energy model based on DC bus voltage fluctuation data; The power reverse derivation unit calculates the charging and discharging power requirements based on the flywheel speed change rate. The torque compensation calculation unit generates the torque adjustment amount of the drive motor through a proportional-integral algorithm.

7. A data center flywheel energy storage system based on an LCL filter according to claim 6, characterized in that: The dynamic stability assessment module includes a vibration spectrum analysis unit, a stability criterion construction unit, and a strategy correction unit. The vibration spectrum analysis unit quantifies and evaluates the mechanical vibration energy of the flywheel rotor; The stability criterion construction unit generates a dynamic stability threshold based on vibration data; The strategy correction unit performs parameter decay correction on control strategies that exceed the threshold.

8. A data center flywheel energy storage system based on an LCL filter according to claim 7, characterized in that: The filter parameter optimization module uses the coupling relationship between flywheel speed feedback data and DC bus voltage to compensate the damping resistance and parallel capacitor parameters of the LCL filter in real time, so as to reduce the harmonic distortion rate of the grid-side current.

Images