Direct current access flywheel energy storage management system for data center

Abstract

The invention relates to the technical field of energy storage management, in particular to a direct current access flywheel energy storage management system for a data center, which comprises a current sequence extraction module, a range signal identification module, a rotating speed inertia calculation module, an output coordination execution module and a control closed loop establishment module. According to the invention, on the basis of continuous current data, a fluctuation trend quantitative model is constructed, range characteristics are superposed to determine a steady-state period, dynamic ratio analysis is carried out by combining a multi-period rotation speed and a current mean value, a response trend measurement index is formed, a power deviation state is matched to execute an adjustment process, and a signal level is adjusted in a linkage manner to construct an operation sequence. A closed-loop control mechanism is realized, and through a whole-process quantitative parameter extraction and response accurate identification mode, the precision of load stability identification is improved, the energy storage response adjustment agility is enhanced, and the power gap compensation response rate and matching precision are improved; and the continuous energy supply capability of the data center in the bus current fluctuation state and the dynamic coordination performance of the flywheel energy storage system are ensured.

Description

Technical Field

[0001] This invention relates to the field of energy storage management technology, and in particular to a DC-connected flywheel energy storage management system for data centers. Background Technology

[0002] The field of energy storage management technology involves the control and optimization of the entire process of energy acquisition, conversion, storage, release, and distribution. It covers dynamic energy management strategies, selection and coordinated control of energy carriers, and collaborative operation technologies with electrical equipment or the power grid. This includes the implementation of energy flow management control logic and the design of safety protection mechanisms for energy storage devices. It is widely used in improving the stability of new energy access to the power system and building energy supply guarantee systems in high-reliability scenarios. Among them, the traditional DC access flywheel energy storage management system for data centers refers to a management system that uses flywheel energy storage devices as backup or transitional energy in scenarios with high requirements for power supply continuity in data centers. It addresses the access control and energy allocation problems of flywheel energy storage devices under DC power supply architecture. The traditional method usually uses a fixed-speed motor to drive the flywheel to rotate. When the mains power is interrupted, the flywheel releases kinetic energy and forms DC power output through the motor reverse drive. At the same time, the connection and disconnection of the power between the flywheel energy storage system and the DC bus of the data center is controlled by hard-wired relay switching. The energy injection rate is regulated by voltage sampling and current-limiting resistor control.

[0003] Existing flywheel energy storage management technology mainly relies on fixed-period voltage sampling for energy regulation and control. Under conditions of continuous small fluctuations or frequent dynamic disturbances in the load current, it is easy to cause lag in the bus voltage response, and relay control actions are prone to misjudgment or delay. The current-limiting resistor has limited adjustment accuracy, resulting in imbalance in the compensation rhythm and inaccurate load status identification during the flywheel energy injection process. When the DC bus is in the critical fluctuation range, it is difficult to effectively determine whether to start the flywheel response process, which may cause a mismatch between the energy release rhythm and the actual gap, thereby affecting the overall power supply continuity of the data center and the operational stability of the energy storage system. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a DC-connected flywheel energy storage management system for data centers.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A DC-connected flywheel energy storage management system for data centers includes: The current sequence extraction module acquires the current sampling sequence within the rolling time window of the DC bus in the data center, establishes a current queue, plots dynamic characteristic curves, and generates a periodic current fluctuation sequence. The range signal identification module calculates the periodic range value based on the periodic current fluctuation sequence, marks the period that meets the steady-state condition, and generates a periodic load fluctuation identification signal. The rotational speed inertia calculation module marks the period of the stable state based on the periodic load fluctuation identification signal, counts the continuous rotational speed data sequence of the flywheel within the corresponding period, calculates the total amplitude of rotational speed change, synchronously extracts the average value of the bus load current within adjacent periods and calculates the current attenuation amplitude, calculates the synchronous ratio of the total amplitude of rotational speed change to the current attenuation amplitude and performs cluster judgment to obtain the energy storage response coordination trend record. The output coordination execution module determines whether the power deviation threshold is lower than the recorded energy storage response coordination trend. If the power deviation threshold is lower than the record, the power gap matching detection is performed to obtain the flywheel energy storage status report. The control closed-loop establishment module reads the power supplement signal level in the adjustment state based on the flywheel energy storage status report, and constructs the operating state sequence by combining the periodic load fluctuation identification signal and the energy storage response trend record, thereby generating a flywheel energy storage management and control scheme.

[0006] As a further embodiment of the present invention, the periodic current fluctuation sequence includes a maximum current value sequence, a minimum current value sequence, and a current fluctuation trajectory map; the periodic load fluctuation identification signal includes a periodic state identification code, a steady-state condition satisfaction identification, and a current fluctuation rate label; the energy storage response coordination trend record includes a rotational speed change ratio, a current attenuation amplitude ratio, and a state synchronization coefficient; the flywheel energy storage status report includes a power deviation result, a response matching status, and a status reliability level; and the flywheel energy storage management and control scheme includes an operating status sequence, a power supplementation level command, and a control execution strategy.

[0007] As a further aspect of the present invention, the current sequence extraction module includes: The data sampling submodule acquires current data within the rolling time window of the DC bus in the data center, extracts and records data changes in the continuous time period, constructs a basic current sample sequence based on the sampling period, and obtains the current sampling sequence. The accuracy queue establishment submodule, based on the current sampling sequence and combined with the standard-defined current measurement accuracy parameters, performs numerical correction and accuracy calibration on the current point values ​​in the sequence, calculates the fluctuation range of each cycle, determines whether each data segment meets the magnetic compatibility measurement level requirements, performs data filtering, and obtains the accuracy calibration current queue. The waveform construction submodule calibrates the current queue according to the accuracy, constructs the dynamic waveform sequence of the current cycle, combines the complete set of sampling points in each cycle, draws the standardized current change curve, extracts the current data feature indicators of each cycle, and generates the cycle current fluctuation sequence.

[0008] As a further aspect of the present invention, the range signal identification module includes: The range calculation submodule collects and processes the point data of each period according to the periodic current fluctuation sequence, calculates the periodic range correction value, calculates the standard offset term according to the sample point distribution within the period, and performs linkage correction with the voltage standard fluctuation amplitude and temperature drift to obtain the periodic range correction value sequence. The steady-state determination submodule performs a steady-state determination operation for each cycle based on the cycle range correction value sequence. It extracts the current fluctuation rate change value in the continuous cycle segment, performs differential processing on the range values ​​of adjacent cycles and calculates their rate of change, counts the average fluctuation value of the rate of change within one minute, compares it with the fluctuation limit, marks the cycle that meets the steady-state requirements, and obtains a steady-state cycle sequence. The fluctuation identifier generation submodule, based on the steady-state periodic sequence, marks each period that meets the conditions as a valid period and adds a fluctuation label, which is then uniformly written into the periodic data structure to establish the fluctuation identifier label data for each period and generate the periodic load fluctuation identifier signal.

[0009] As a further aspect of the present invention, the rotational speed inertia calculation module includes: The speed fluctuation extraction submodule reads the corresponding flywheel speed sequence data based on the period marked as a stable state by the periodic load fluctuation identification signal, extracts all speed point values ​​in each period sampling sequence, sorts them according to time order, and calculates the total change in flywheel speed in each period to generate a speed change amplitude sequence. The current decay calculation submodule, based on the periodic load fluctuation identification signal marked as a stable period, sequentially obtains the bus load current sequence of each pair of adjacent periods, combines it with the speed change amplitude sequence, calculates the joint change rate of the total speed change amplitude and the current decay amplitude, establishes the mapping relationship between the change rate and the current period number, and obtains the inertial response amplitude ratio sequence. The response coordination trend generation submodule reads the corresponding ratio for each cycle according to the inertial response amplitude ratio sequence and arranges them in time sequence. It calculates the ratio change rate for the changes in ratios of adjacent cycles, statistically analyzes the overall increase or decrease trend of the ratio change rate sequence, and performs clustering judgment based on the trend of change to obtain the energy storage response coordination trend record.

[0010] As a further aspect of the present invention, the power output coordination execution module includes: The response trend judgment submodule determines whether the response trend of the current period is continuously decreasing based on the energy storage response and trend record. If the change amplitude is less than 0 in three consecutive periods and the change amount exceeds the judgment interval each time, the current period is marked as a response capability decay state. The response trend value of the last period in the trend record is compared with the set trend coefficient benchmark value. If it is lower than the benchmark value, it is confirmed that the response trend is in the declining segment, and the response decay identification result is obtained. Based on the response attenuation identification result, the deviation trigger identification submodule combines the current cycle bus output power sampling data and flywheel response power recording data to perform difference calculation on the two data and obtain the absolute difference. It then performs normalization conversion with the rated power value as a reference, and calculates the power deviation percentage under the corresponding cycle. If it is greater than the power deviation threshold, the corresponding cycle is marked as a deviation activation state, and the power difference state analysis record is obtained. The energy storage status extraction submodule reads the response time record and power response error data of the flywheel system in the current cycle based on the power difference status analysis record, determines the status mark of the current cycle, analyzes the distribution trend of the flywheel status in each cycle, and classifies and determines the flywheel energy storage status report based on the distribution trend.

[0011] As a further aspect of the present invention, the control closed-loop establishment module includes: Based on the flywheel energy storage status report, the power level extraction submodule reads the power supplementation signal in the corresponding period's adjustment state, sets the power supplementation reference value, determines the period's supplementation state, records the corresponding supplementation level identifier, and obtains the power supplementation signal level value. The state sequence construction submodule reads the periodic load fluctuation identification signal and the energy storage response trend record based on the power supplementation signal level value, and constructs a state triplet by combining the power supplementation level, load fluctuation amplitude and response trend. After sorting each state triplet by period number, a unified state structure list is constructed to establish the operating state sequence. The strategy scheme generation submodule extracts the cycle number and numerical information corresponding to each state triplet based on the operating state sequence, identifies the degree of deviation of the cycle in terms of load fluctuation amplitude and trend value, sets a stable threshold range, marks the active segment of state change, maps and organizes the data according to the cycle, and obtains the flywheel energy storage management and control scheme.

[0012] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, a fluctuation trend quantification model is constructed based on continuous current data. The steady-state period is determined by superimposing range characteristics. Dynamic ratio analysis is performed by combining multi-cycle rotational speed and current average to form a response trend measurement index. The power deviation state is matched to execute the adjustment process. The linkage adjustment signal level is used to construct the operation sequence, realizing a closed-loop control mechanism for energy storage output. Through the extraction of quantitative parameters and accurate response identification in the whole process, the accuracy of load stability identification is improved, the agility of energy storage response adjustment is enhanced, the response rate and matching accuracy of power gap compensation are improved, and the continuous power supply capability of the data center under the bus current fluctuation state and the dynamic coordination performance of the flywheel energy storage system are guaranteed. Attached Figure Description

[0013] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the current sequence extraction module of the present invention; Figure 3 This is a flowchart of the range signal identification module of the present invention; Figure 4 This is a flowchart of the rotational speed and inertia calculation module of the present invention; Figure 5 This is a flowchart of the power coordination and execution module of the present invention; Figure 6 This is a flowchart of the control closed-loop establishment module of the present invention. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0015] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0016] Please see Figure 1 A DC-connected flywheel energy storage management system for data centers includes: The current sequence extraction module acquires the current sampling sequence within the rolling time window of the DC bus of the data center (a 10ms sampling period defined by the safety standard for electrical drive systems), establishes a Class A current queue (the highest level standard for magnetic compatibility measurement, with current measurement accuracy of ±0.1% of the reading and ±0.05% of the range, and a storage depth of ≥10 periods), plots dynamic characteristic curves (voltage / current dynamic recording format defined by the power quality monitoring standard, containing waveform data of 256 sampling points / period), and generates a periodic current fluctuation sequence. The range signal identification module calculates the periodic range value based on the periodic current fluctuation sequence, compares the periodic range value with the voltage fluctuation limit, marks the period that meets the steady-state conditions (steady-state judgment conditions defined by the electromagnetic compatibility standard: current fluctuation rate ≤3% / minute, duration ≥30 seconds), and generates a periodic load fluctuation identification signal; The rotational speed inertia calculation module is based on the periodic load fluctuation identification signal marked as the stable state period. It statistically analyzes the continuous rotational speed data sequence of the flywheel within the corresponding period, calculates the total amplitude of rotational speed change, synchronously extracts the average value of the bus load current in adjacent periods and calculates the current attenuation amplitude, calculates the synchronous ratio of the total amplitude of rotational speed change to the current attenuation amplitude, and performs clustering judgment based on the trend of change to obtain the energy storage response coordination trend record. The output coordination execution module determines whether the power deviation threshold (grid-connected power generation equipment standard: power deviation ≤ 5% of rated power) is lower than the power deviation threshold based on the energy storage response and coordination trend record. If it is lower than the threshold, it performs power gap matching detection (response time ≤ 200ms, power tracking error ≤ 3%) and obtains a flywheel energy storage status report. The control closed-loop establishment module is based on the flywheel energy storage status report, reads the power supplement signal level in the adjustment state, and constructs the operating status sequence by combining the periodic load fluctuation identification signal and the energy storage response trend record, thereby generating a flywheel energy storage management and control scheme.

[0017] The periodic current fluctuation sequence includes the maximum current value sequence, the minimum current value sequence, and the current fluctuation trajectory map. The periodic load fluctuation identification signal includes the periodic status identification code, the steady-state condition satisfaction indicator, and the current fluctuation rate label. The energy storage response coordination trend record includes the speed change ratio, the current attenuation amplitude ratio, and the state synchronization coefficient. The flywheel energy storage status report includes the power deviation result, the response matching status, and the status reliability level. The flywheel energy storage management and control scheme includes the operating status sequence, the power supplementation level command, and the control execution strategy.

[0018] Please see Figure 2 The current sequence extraction module includes: The data sampling submodule acquires current data within the rolling time window of the DC bus in the data center, extracts and records data changes in the continuous time period, constructs a basic current sample sequence based on the sampling period, and obtains the current sampling sequence. To obtain current data from the DC bus within a rolling time window in a data center, a bus topology monitoring point map needs to be constructed based on the data center server backbone power supply structure. Current sensors should be deployed at typical DC nodes, and a rolling detection window should be set up for continuous sampling at 10ms intervals. For example, if the sampling start point is set to 0ms, current data points will be continuously acquired at 10ms, 20ms, 30ms, and so on, up to the previous cycle, forming a continuous sampling data series. The sampling points should use an AD conversion module to read analog signals, and synchronous acquisition should be controlled by a built-in clock. 256 sampling points should be acquired per cycle, i.e., 25,600 data points per second. This sample size meets the minimum waveform resolution requirement for power quality monitoring. Subsequently, the current values ​​at each sampling point are normalized for the next stage of calibration and comparison. For example, for a... If the sequence collected by each node is [22.4A, 22.6A, 23.1A, ...], then it should be normalized point by point to limit it to the range of 0~1. Using the system's default range setting of ±100A, the normalized sequence is [(22.4+100) / 200, (22.6+100) / 200, ...]. The above normalization process provides the basis for subsequent accuracy standard comparison. The sampling sequence needs to be cached in a temporary register group as a byte stream, and a sequence number and timestamp should be added according to the node identifier. At the same time, the start and end times of each sampling segment should be recorded to control the sliding window update strategy. For example, if the sliding window width is 500ms, the sliding window boundary should be updated every 50 sets of data sampled, and the old data should be moved out of the cache to ensure storage efficiency and real-time performance. Finally, a continuous dynamic sampling data sequence is established to obtain the current sampling sequence.

[0019] The accuracy queue establishment submodule is based on the current sampling sequence. Combined with the current measurement accuracy parameters defined by the standard, it performs numerical correction and accuracy calibration on the current point values ​​in the sequence, calculates the fluctuation range of each cycle, determines whether each data segment meets the magnetic compatibility measurement level requirements, performs data filtering, and obtains the accuracy calibration current queue. Based on the current sampling sequence, measurement error correction processing is required for the current values ​​at each sampling point. The error requirements of ±0.1% of the reading and ±0.05% of the range in the Class A current measurement standard are invoked. Error correction matrices are constructed using single-point calibration deviation and overall measurement range offset, respectively. Double correction is applied to each point value in the current sequence, i.e., subtracting both its theoretical value deviation and the overall range lower limit offset. For example, if a sampling point reading is 23.5A, the theoretical value is 23.3A, and the measurement error is 0.2A, and the range is set to ±100A, then the ±0.05% range deviation would be 0.1A, and the total error would be... The upper limit is 0.3A. A correction operation is required to calibrate the value to 23.3A before it is included in the queue construction. Subsequently, in a sampling window of 10 cycles, the average, maximum, minimum, and variation range of 256 data points in each cycle are statistically analyzed, and the absolute offset value of each point from the cycle average is calculated. If the fluctuation range of this offset value is less than a set threshold in multiple consecutive cycles, the data segment is determined to be a valid segment. This threshold is the fluctuation tolerance range set according to the electromagnetic compatibility standard, which is set to 0.25A. Refer to the sampling data results listed in Table 1. All consecutive cycle segments that meet the conditions are selected for constructing the accuracy queue. Table 1: Comparison of Fluctuations in Sampled Data .

[0020] As shown in Table 1, periods 2 and 3 meet the fluctuation tolerance requirements and can be included in the Class A precision current queue, ultimately resulting in the precision calibration current queue.

[0021] The waveform construction submodule calibrates the current queue based on accuracy, constructs the dynamic waveform sequence of the current cycle, combines the complete set of sampling points in each cycle, draws the standardized current change curve, extracts the current data characteristic indicators of each cycle, and generates the cycle current fluctuation sequence. Based on the periodic sampling data in the precision calibration current queue, a dynamic waveform sequence of 256 points / cycle is constructed. First, taking each cycle as a unit, the 256 sampling points of each cycle are grouped into a complete array to form a waveform dataset. Then, all periodic waveforms are arranged along the time axis to form a two-dimensional current fluctuation matrix. The matrix dimension is the number of cycles × 256. For example, if 8 consecutive cycles are selected, an 8×256 data table is formed. This matrix is ​​used to plot the two-dimensional curve of current changing with time, with the horizontal axis representing the sampling point index and the vertical axis representing the current value. For each waveform curve, the maximum, minimum, and average values ​​are extracted as periodic feature quantities. These represent the intensity range and central trend of the current change in the cycle, respectively. Then, the characteristic difference between adjacent cycles is calculated, and the trend changes are compared to determine the fluctuation pattern. In specific implementation, timestamps need to be added to the sampled data to ensure the accuracy of cycle division. For example, the first cycle starts at 1000ms, and 256 points are sampled up to 1025.6ms. The second cycle starts sampling from 1025.6ms to ensure waveform continuity. Finally, the waveform and feature array corresponding to each cycle are stored sequentially to form a complete fluctuation database for subsequent dynamic characteristic evaluation and backtracking query, ultimately generating a cycle current fluctuation sequence.

[0022] Please see Figure 3 The range signal identification module includes: The range calculation submodule aggregates and processes the point data for each period based on the periodic current fluctuation sequence, using the following formula: ; The periodic range correction value is calculated. Based on the sample point distribution within the period, the standard offset term is calculated. This is then combined with a linkage correction based on the voltage standard fluctuation amplitude and temperature drift, resulting in a sequence of periodic range correction values. This is the correction value for the periodic range after correction. Let be the current value at the i-th sampling point, in A. Let t be the set of points in periodicity t. The mean current in period t, in A. The standard deviation of voltage fluctuation, in V. This is the voltage reference value, in volts (V). This is the temperature value at the current sampling point, in °C. The ambient reference temperature is expressed in °C. Based on the periodic current fluctuation sequence, it is necessary to extract the maximum and minimum values ​​for each period. The first step is to construct a set of sampling points within the period. Each group contains 256 data points. The extraction method is based on time-series index truncation, automatically dividing the data into blocks according to period numbering, and then calling them separately. and The function retrieves the upper and lower limits of the period. Taking period 1 as an example, its maximum value is 28.6A and its minimum value is 22.1A, a difference of 6.5A. This is then combined with the standard deviation of the period voltage. With reference voltage The ratio is adjusted for fluctuation compensation. Taking period 1 as an example, the standard deviation is 0.8V, the reference voltage is 220V, and the correction ratio is... At the same time, obtain the current temperature Reference temperature Temperature offset ratio is Therefore, the correction term is Then, combining the square root of the sum of the mean of the sampling points within the period and the absolute value of the deviation of each sampling point (i.e., the mean square term of fluctuation), let the value corresponding to period 1 be... Substitute the values ​​into the formula to perform the calculation: ; Calculate period 2 and period 3 using the same method: ; ; Table 2. Calculation Parameters for Periodic Range Correction Values ​​(Units: Current (A), Voltage (V), Temperature (°C)). .

[0023] As shown in Table 2, the three sets of periodic samples form corrected periodic range values ​​after temperature and voltage compensation, which provide a basis for subsequent steady-state determination and finally obtain the periodic range correction value sequence.

[0024] The period range correction value is a composite numerical index used to quantify the fluctuation intensity of current within a single sampling period. It not only reflects the difference in the original current amplitude between the maximum and minimum current values ​​within the period, but also introduces environmental factors such as voltage fluctuations and temperature deviations for dynamic correction. This makes the value sensitive and adaptable to external interference sources, and can more realistically depict the load stability or sudden change characteristics of the power system under a specific period. Its positive or negative sign can be used to distinguish the direction of the fluctuation trend after correction, and its numerical value is used to measure whether the fluctuation intensity is in the deviation abnormality or steady state range in the current period. It is the core quantitative basis for subsequent steady state identification and fluctuation identification judgment in the subsequent period.

[0025] The formula's operational logic is based on multiple corrections to the periodic current range, forming a fluctuating scalar with environmental adaptability. Firstly, the first term of the formula is the periodic current range. This is used to quantify the maximum current fluctuation amplitude within a cycle; this difference serves as a basis for reflecting the severity of changes in the electrical load. Secondly, a temperature compensation term and a voltage compensation term are introduced to form a multiplication factor within parentheses. In this term, the two ratios represent the degree of periodic voltage fluctuation and the degree of temperature deviation, respectively. The subtraction operation reflects the consistency or cancellation relationship of the direction of environmental disturbances. When the voltage fluctuation and temperature deviation are in opposite directions, the correction amplitude can be reduced, enhancing the sensitivity to identifying stable periods. Finally, multiply by the inverse root mean square term. This mechanism is used to constrain the excessive amplification of range correction when the sampling value distribution width within the period is large. Its function is to normalize the current dispersion within the period, thereby forming a composite discrimination structure that links fluctuation intensity, environmental correction, and internal equilibrium, reflecting a multi-factor fusion current range correction mechanism. The overall calculation uses multiplication to construct the mutual trade-off between different dimensional factors, and reduces the outlier pull effect by taking the square root, ultimately forming a quantifiable and comparable periodic fluctuation index.

[0026] The steady-state determination submodule is based on the periodic range correction value sequence. It performs steady-state determination operation for each period, extracts the current fluctuation rate change value in the continuous period segment, performs differential processing on the range values ​​of adjacent periods and calculates their rate of change, counts the average fluctuation value of the rate of change within one minute, compares it with the fluctuation limit, marks the period that meets the steady-state requirements, and obtains the steady-state periodic sequence. After obtaining the periodic range correction value sequence, it is necessary to calculate the rate of change of the difference between adjacent periods for each period to determine whether the steady-state condition is met. Assuming that a one-minute window contains 120 periods, the correction value is selected. and By performing term-by-term difference calculation, modulo averaging, and defining volatility as... If the value is not greater than 0.03 and the duration of the period that continuously meets this standard is greater than 30 seconds (i.e., 60 periods), then this time period is determined to be a steady-state segment. For example, if the correction values ​​for periods 1, 2, and 3 are -1.010, -0.676, and -0.514 respectively, then the rates of change are respectively: ; ; In the above results, the first pair of periodic changes is 0.334, which exceeds the judgment threshold and does not meet the condition. The second pair of periodic changes is 0.162, which is less than 0.03 and meets the condition, but does not meet the requirement of 60 consecutive periods. Therefore, the current judgment period segment does not constitute a steady-state segment. In actual engineering, a sliding statistical judgment window needs to be introduced to perform this judgment process on the complete sample set, and finally obtain a steady-state periodic sequence.

[0027] The fluctuation identifier generation submodule is based on a steady-state periodic sequence. It marks each period that meets the conditions as a valid period and adds a fluctuation label. These are then uniformly written into the periodic data structure to establish the fluctuation identifier label data for each period and generate the periodic load fluctuation identifier signal. Based on the steady-state cycle sequence, a label assignment operation is performed on the cycle number that meets the steady-state judgment condition. The system has a preset "cycle state flag" field Flag, and each cycle records this field independently. If the number exists in the aforementioned steady-state sequence, the flag is set to 1, otherwise it is 0. For example, if cycles 4 to 64 all meet the steady-state change condition, then the Flag for cycles 4 to 64 is marked as 1, and the other cycles are marked as 0. The marking results are synchronously written into the load record structure, sorted by timestamp to form a current fluctuation identification chain, and combined with time, sampling number, and current value to form a cycle state object array, finally generating a cycle load fluctuation identification signal.

[0028] Please see Figure 4 The rotational speed and inertia calculation module includes: The speed fluctuation extraction submodule reads the corresponding flywheel speed sequence data based on the period marked as a stable state by the periodic load fluctuation identification signal, extracts all speed point values ​​in the sampling sequence of each period, sorts them in time order, and calculates the total change in flywheel speed in each period to generate a speed change amplitude sequence. When extracting the flywheel speed sequence based on the periods marked as stable states, it is necessary to first identify the number of each marked period. Taking period 1 as an example, its sampled speed point value sequence is as follows: After arranging the time sequence, perform a difference operation on any two adjacent speed points, and calculate sequentially: , , Then, take the absolute value of the set of differences and sum them to obtain the sum of the speed fluctuation terms for period 1. Then, the mean value of the rotational speed points in that cycle is calculated as follows: Then its average fluctuation range is The sequence is obtained in period 2 in the same way. The difference is Summing gives 19, with a mean of 1542.75 rad / s. The sequence in period 3 is... The fluctuation difference is also 19, with a mean of 1520.75 rad / s. By performing differential, normalization, and mean statistical processing on the flywheel speed point sequence within multiple stable periods, a speed fluctuation amplitude index within the period can be formed. This index will be used as a basic quantity for the dynamic inertia change of the flywheel in subsequent calculations. The specific data is listed below: Table 3 Calculation of rotational speed and current attenuation parameters: .

[0029] As shown in Table 3, a set of mean speed and fluctuation terms for the corresponding period has been formed, and finally a sequence of speed change amplitude is generated.

[0030] The current decay calculation submodule, based on the periodic load fluctuation identification signal marking the period as a stable state, sequentially obtains the bus load current sequence for each pair of adjacent periods, and combines it with the speed change amplitude sequence, using the formula: ; Calculate the joint rate of change of the total rotational speed change and the current decay rate, establish a mapping relationship between the rate of change and the current cycle number, and obtain the inertial response amplitude ratio sequence, where, Indicates the first The periodic energy storage response and trend coefficient are dimensionless proportional values. For the first The first in the cycle The flywheel speed at a given moment, in units of ; This is the arithmetic mean of all rotational speeds during this period, in units of... ; This represents the number of rotational speed samples. , These are the average bus current values ​​for the current cycle and the previous cycle, respectively, in units of... ; This represents the total number of flywheel speed samples within the current cycle. For the first Each rotational speed point value; Based on the extracted stable cycle information, the bus load current data for each pair of adjacent cycles is retrieved. Taking cycle 1 and cycle 2 as examples, the average current in the preceding cycle (cycle 1) is 65.2A, and the average current in the following cycle (cycle 2) is 64.1A. The difference is calculated as follows: A, the relative attenuation magnitude is The flywheel speed sequence for period 2 is as follows: Its point mean is 1542.75 rad / s, and the total difference is 19. Substituting these values ​​into the formula, the normalized difference is... Continue calculating its rotational speed fluctuation dispersion term: The relative distance of each point from the mean is calculated as follows: ; ; ; ; The average value is The denominator is ; The final calculated joint rate of change for this period is: ; When repeating the operation cycle 3, the joint rate of change can be obtained using the same calculation method. This result indicates that the flywheel's normalized speed response change has moderate synchronicity relative to the load current decay amplitude and its own dispersion, thus obtaining the inertial response amplitude ratio sequence.

[0031] The combined rate of change is a comprehensive dimensionless index used to measure the degree of coordination between the internal speed change of the flywheel system and the external load current decay within the same cycle. It reflects whether the two show a synchronous or mismatched trend in time dynamics by analyzing the ratio of the normalized difference of the flywheel speed sampling points within the cycle to the difference of the average bus current in adjacent cycles. When the combined rate of change is high, it indicates that the rapid response of the flywheel speed and the decreasing trend of the load current are significantly linked. When the ratio is low, it indicates that the response deviation between the two is large, and an effective inertial support coordination cannot be formed. Therefore, this value can be used to evaluate the flywheel's inertial response capability to current decay changes under stable cycles, and serve as a quantitative basis for subsequent control regulation or system judgment.

[0032] The formula's operational logic is based on a joint normalized structure between speed change and current decay, aiming to construct a dimensionless dynamic trend indicator. The numerator accumulates all speed differences within a cycle through summation, dividing each difference by the current cycle's average speed to obtain multiple normalized speed change rates. These absolute values ​​are then summed to reflect the overall trend, ensuring that positive and negative fluctuations do not cancel each other out. The first term in the denominator is the difference between the current cycle's average current and the previous cycle's average current, forming a relative current decay ratio to measure the external response intensity of load fluctuations. The second term in the denominator is the sum and average of the relative deviations of all speed points from the mean (i.e., standardized deviations), characterizing the speed dispersion within the cycle; this reflects the internal dynamic fluctuation background. The overall denominator is a fusion of external decay and internal fluctuation factors, combined through addition to form a normalized reference term. This term is divided by the normalized speed change value in the numerator to form a trend ratio, ensuring numerical logical symmetry and dimensional consistency, thus forming a stable trend quantification indicator for cross-cycle comparison.

[0033] The response coordination trend generation submodule reads the corresponding ratio of each cycle according to the inertial response amplitude ratio sequence and arranges them in time sequence. It calculates the ratio change rate for the ratio changes in adjacent cycles, statistically analyzes the overall increase and decrease trend of the ratio change rate sequence, and performs clustering judgment based on the change trend to obtain the energy storage response coordination trend record. Based on the obtained inertial response amplitude ratio sequence, the corresponding coefficients for each period are sorted and combined sequentially, as shown in Table 3: Period 1 = 1.10, Period 2 = 1.07, Period 3 = 1.03. Differences are then performed sequentially: the difference between Period 2 and Period 1 is... The difference between period 3 and period 2 is Since both differences are below the clustering difference window of 0.05, the three periods are grouped into the same response trend category, and the mean within this group is calculated as follows: This is considered the representative trend center of the sequence, used to identify the numerical value of the coordination state between the flywheel and the load during this period, and finally obtains the energy storage response coordination trend record.

[0034] Please see Figure 5 The output coordination and execution module includes: The response trend judgment submodule determines whether the response trend of the current period is continuously declining based on the energy storage response and trend record. If the change amplitude is less than 0 in three consecutive periods and the change amount exceeds the judgment interval each time, the current period is marked as a response capability decay state. The response trend value of the last period in the trend record is compared with the set trend coefficient benchmark value. If it is lower than the benchmark value, it is confirmed that the response trend is in the declining segment, and the response decay identification result is obtained. Based on the trend coefficient corresponding to each cycle in the energy storage response coordination trend record, the response trend values ​​for cycles 1, 2, and 3 were extracted as 0.97, 0.93, and 0.90, respectively. The changes during the cycles were calculated sequentially. The change in cycle 2 compared to cycle 1 was -0.04, the change in cycle 3 compared to cycle 2 was -0.03, and the change in cycle 3 compared to cycle 1 was -0.07. The changes in all three cycles were less than 0, and the minimum change during the cycle reached -0.02, indicating that the current trend change has a continuous downward characteristic. Subsequently, the trend coefficient value of 0.90 for cycle 3 was extracted and compared with the trend coefficient benchmark value of 0.95. Since 0.90 is lower than the benchmark value of 0.95, it was determined that the trend is in the downward range. Based on the number and magnitude of the downward trend, it was determined that the response capability of the current cycle has weakened.

[0035] Table 4 Data Table for Energy Storage Response Coordination Module: .

[0036] As shown in Table 4, the periodic trend coefficient decreases continuously and the magnitude exceeds the set judgment standard. The trend value corresponding to period 3 is lower than the benchmark value, thus obtaining the response decay identification result.

[0037] The deviation trigger identification submodule, based on the response decay identification result, combines the current cycle bus output power sampling data and flywheel response power recording data, performs difference calculation on the two data and obtains the absolute difference, performs normalization conversion with the rated power value as a reference, and calculates the power deviation percentage under the corresponding cycle. If it is greater than the power deviation threshold, the corresponding cycle is marked as deviation activation state, and the power difference state analysis record is obtained. Based on the response attenuation identification results, the flywheel response power and bus power of cycles numbered 1 to 3 are used to calculate the power deviation. In cycle 1, the bus power is 100.0kW, the flywheel response is 94.0kW, and the deviation is 6.0kW, accounting for 6.0% of the rated power of 100.0kW. In cycle 2, the deviation is 5.8kW, accounting for 5.8%, and in cycle 3, the deviation is 4.9kW, accounting for 4.9%. Using the power deviation threshold of 5% as the comparison standard, it is determined that the deviations of cycle 1 and cycle 2 exceed the threshold, indicating a deviation activation state, while the deviation of cycle 3 does not exceed the threshold, indicating a deviation inactive state. The corresponding cycle number and state value are recorded in the deviation trigger flag table, marking the state value of cycle 1 and cycle 2 as 1 and the state value of cycle 3 as 0, respectively, and finally obtaining the power difference state analysis record.

[0038] The energy storage status extraction submodule reads the response time record and power response error data of the flywheel system in the current cycle based on the power difference status analysis record, determines the status mark of the current cycle, analyzes the distribution trend of the flywheel status in each cycle, and classifies and judges the status based on the distribution trend to obtain the flywheel energy storage status report. Based on the power difference state analysis records, the response time and power error values ​​recorded in cycles 1 to 3 were used as the judgment criteria. The response time of cycle 1 was 185ms and the power error was 2.8%. The response time of cycle 2 was 198ms and the power error was 2.9%. Both met the conditions that the response time should not exceed 200ms and the power error should not exceed 3%. The response time of cycle 3 was 207ms and the power error was 3.2%. Neither met the judgment conditions. The state flags of each cycle were assigned values. Cycle 1 and cycle 2 were marked as 1, and cycle 3 was marked as 0. Further statistics showed that the number of times the state value was 1 in each cycle was 2, accounting for 66.67%. This was recorded as the state distribution trend, and finally the flywheel energy storage state report was obtained.

[0039] Please see Figure 6 The control closed-loop establishment module includes: The power level extraction submodule reads the power supplement signal in the corresponding period's adjustment state based on the flywheel energy storage status report, sets the power supplement reference value, determines the period's supplement status, records the corresponding supplement level identifier, and obtains the power supplement signal level value. Based on the flywheel energy storage status report, data for cycles numbered 1 to 3 with a status mark of 1 were obtained, namely cycle 1 and cycle 2. Cycle 3, with a status mark of 0, was not included in the analysis. The power supplement signal for cycle 1 was extracted as 6.0kW, and for cycle 2 as 5.8kW. The power supplement benchmark value was set to 5.0kW, which was determined by the preset experience of the percentage of rated power. This threshold was set with reference to the standard response behavior of 10 systems with a rated power of 100kW to determine the supplement intensity in the power response stage. The supplement signals for cycles 1 and 2 were both greater than 5.0kW, and were determined to be supplement level 2. The supplement level setting rules are as follows: less than or equal to 5.0kW is set as level 1, 5.0kW to 7.5kW is set as level 2, and higher than 7.5kW is set as level 3. Cycle 3 was not included in the level judgment because its status mark was 0. The supplement signal level values ​​were calculated and marked as 2 for cycle 1, 2 for cycle 2, and 1 for cycle 3, respectively. The results are shown in Table 5. Finally, the power supplement signal level values ​​were obtained. Table 5 Energy Storage Control Status Parameters: .

[0040] As shown in Table 5, both period 1 and period 2 constitute moderate supplementary responses, while period 3 does not meet the state activation condition.

[0041] The state sequence construction submodule reads the trend record of the periodic load fluctuation identification signal and the energy storage response based on the power supplementation signal level value. It constructs a state triplet by combining the power supplementation level, load fluctuation amplitude and response trend. After sorting each state triplet by period number, it constructs a unified state structure list and establishes the operating state sequence. Based on the power supplement signal level value, the load fluctuation amplitude and response trend value of cycle 1 and cycle 2 are used to construct a state triplet structure. The three data of cycle 1 are {2, 3.4, 0.97}, and those of cycle 2 are {2, 4.1, 0.93}. The two sets of triplets are used to construct a sorted state sequence according to the cycle number and to encode the state structure. In the state offset marker, cycle 1 is assigned the value "high" because the fluctuation amplitude is relatively small and the trend value is high. Cycle 2 is assigned the value "medium" because the fluctuation amplitude is further increased and the trend value is decreased. Cycle 3 is marked as 0 for energy storage state. Although it has a fluctuation amplitude of 3.0% and a trend value of 0.90, it does not enter the state sequence structure and is only used to assist in background comparison. Finally, the operating state sequence is established.

[0042] The strategy scheme generation submodule extracts the cycle number and numerical information corresponding to each state triplet based on the operating state sequence, identifies the degree of deviation of the cycle in load fluctuation amplitude and trend value, sets a stable threshold range, marks the active segment of state change, maps and organizes by cycle, and obtains the flywheel energy storage management and control scheme. Based on the operating state sequence, the state triplets constructed in cycle 1 and cycle 2 are read, and their load fluctuation amplitude and trend value are compared item by item. The load fluctuation in cycle 1 is 3.4% and the trend value is 0.97, while the load fluctuation in cycle 2 is 4.1% and the trend value is 0.93. The fluctuation amplitude increases by 0.7% and the trend value decreases by 0.04. It is determined that the fluctuation is stronger and the trend is weaker during consecutive cycles. The trend judgment threshold is set to 0.05, which is obtained through the evaluation of energy storage operation stability experiment. Based on the fact that the fluctuation trend of each cycle is moving in a negative direction, it is determined that the current cycle is in a downward offset state. The control flag value is marked as -1 for cycle 1 and cycle 2, and marked as 0 for cycle 3. The control information is synchronously written into the cycle control structure table, and finally the flywheel energy storage management and control scheme is obtained.

[0043] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A DC-connected flywheel energy storage management system for data centers, characterized in that, The system includes: The current sequence extraction module acquires the current sampling sequence within the rolling time window of the DC bus in the data center, establishes a current queue, plots dynamic characteristic curves, and generates a periodic current fluctuation sequence. The range signal identification module calculates the periodic range value based on the periodic current fluctuation sequence, marks the period that meets the steady-state condition, and generates a periodic load fluctuation identification signal. The rotational speed inertia calculation module marks the period of the stable state based on the periodic load fluctuation identification signal, counts the continuous rotational speed data sequence of the flywheel within the corresponding period, calculates the total amplitude of rotational speed change, synchronously extracts the average value of the bus load current within adjacent periods and calculates the current attenuation amplitude, calculates the synchronous ratio of the total amplitude of rotational speed change to the current attenuation amplitude and performs cluster judgment to obtain the energy storage response coordination trend record. The output coordination execution module determines whether the power deviation threshold is lower than the recorded energy storage response coordination trend. If the power deviation threshold is lower than the record, the power gap matching detection is performed to obtain the flywheel energy storage status report. The control closed-loop establishment module reads the power supplement signal level in the adjustment state based on the flywheel energy storage status report, and constructs the operating state sequence by combining the periodic load fluctuation identification signal and the energy storage response trend record, thereby generating a flywheel energy storage management and control scheme.

2. The DC-connected flywheel energy storage management system for data centers according to claim 1, characterized in that, The periodic current fluctuation sequence includes a maximum current value sequence, a minimum current value sequence, and a current fluctuation trajectory map. The periodic load fluctuation identification signal includes a periodic state identification code, a steady-state condition satisfaction indicator, and a current fluctuation rate label. The energy storage response coordination trend record includes a rotational speed change ratio, a current attenuation amplitude ratio, and a state synchronization coefficient. The flywheel energy storage status report includes a power deviation result, a response matching status, and a status reliability level. The flywheel energy storage management and control scheme includes an operating status sequence, a power supplementation level command, and a control execution strategy.

3. The DC-connected flywheel energy storage management system for data centers according to claim 1, characterized in that, The current sequence extraction module includes: The data sampling submodule acquires current data within the rolling time window of the DC bus in the data center, extracts and records data changes in the continuous time period, constructs a basic current sample sequence based on the sampling period, and obtains the current sampling sequence. The accuracy queue establishment submodule, based on the current sampling sequence and combined with the standard-defined current measurement accuracy parameters, performs numerical correction and accuracy calibration on the current point values ​​in the sequence, calculates the fluctuation range of each cycle, determines whether each data segment meets the magnetic compatibility measurement level requirements, performs data filtering, and obtains the accuracy calibration current queue. The waveform construction submodule calibrates the current queue according to the accuracy, constructs the dynamic waveform sequence of the current cycle, combines the complete set of sampling points in each cycle, draws the standardized current change curve, extracts the current data feature indicators of each cycle, and generates the cycle current fluctuation sequence.

4. The DC-connected flywheel energy storage management system for data centers according to claim 1, characterized in that, The range signal identification module includes: The range calculation submodule collects and processes the point data of each period according to the periodic current fluctuation sequence, calculates the periodic range correction value, calculates the standard offset term according to the sample point distribution within the period, and performs linkage correction with the voltage standard fluctuation amplitude and temperature drift to obtain the periodic range correction value sequence. The steady-state determination submodule performs a steady-state determination operation for each cycle based on the cycle range correction value sequence. It extracts the current fluctuation rate change value in the continuous cycle segment, performs differential processing on the range values ​​of adjacent cycles and calculates their rate of change, counts the average fluctuation value of the rate of change within one minute, compares it with the fluctuation limit, marks the cycle that meets the steady-state requirements, and obtains a steady-state cycle sequence. The fluctuation identifier generation submodule, based on the steady-state periodic sequence, marks each period that meets the conditions as a valid period and adds a fluctuation label, which is then uniformly written into the periodic data structure to establish the fluctuation identifier label data for each period and generate the periodic load fluctuation identifier signal.

5. The DC-connected flywheel energy storage management system for data centers according to claim 1, characterized in that, The rotational speed inertia calculation module includes: The speed fluctuation extraction submodule reads the corresponding flywheel speed sequence data based on the period marked as a stable state by the periodic load fluctuation identification signal, extracts all speed point values ​​in each period sampling sequence, sorts them according to time order, and calculates the total change in flywheel speed in each period to generate a speed change amplitude sequence. The current decay calculation submodule, based on the periodic load fluctuation identification signal marked as a stable period, sequentially obtains the bus load current sequence of each pair of adjacent periods, combines it with the speed change amplitude sequence, calculates the joint change rate of the total speed change amplitude and the current decay amplitude, establishes the mapping relationship between the change rate and the current period number, and obtains the inertial response amplitude ratio sequence. The response coordination trend generation submodule reads the corresponding ratio for each cycle according to the inertial response amplitude ratio sequence and arranges them in time sequence. It calculates the ratio change rate for the changes in ratios of adjacent cycles, statistically analyzes the overall increase or decrease trend of the ratio change rate sequence, and performs clustering judgment based on the trend of change to obtain the energy storage response coordination trend record.

6. The DC-connected flywheel energy storage management system for data centers according to claim 1, characterized in that, The output coordination execution module includes: The response trend judgment submodule determines whether the response trend of the current period is continuously decreasing based on the energy storage response and trend record. If the change amplitude is less than 0 in three consecutive periods and the change amount exceeds the judgment interval each time, the current period is marked as a response capability decay state. The response trend value of the last period in the trend record is compared with the set trend coefficient benchmark value. If it is lower than the benchmark value, it is confirmed that the response trend is in the declining segment, and the response decay identification result is obtained. Based on the response attenuation identification result, the deviation trigger identification submodule combines the current cycle bus output power sampling data and flywheel response power recording data to perform difference calculation on the two data and obtain the absolute difference. It then performs normalization conversion with the rated power value as a reference, and calculates the power deviation percentage under the corresponding cycle. If it is greater than the power deviation threshold, the corresponding cycle is marked as a deviation activation state, and the power difference state analysis record is obtained. The energy storage status extraction submodule reads the response time record and power response error data of the flywheel system in the current cycle based on the power difference status analysis record, determines the status mark of the current cycle, analyzes the distribution trend of the flywheel status in each cycle, and classifies and determines the flywheel energy storage status report based on the distribution trend.

7. The DC-connected flywheel energy storage management system for data centers according to claim 1, characterized in that, The control closed-loop establishment module includes: Based on the flywheel energy storage status report, the power level extraction submodule reads the power supplementation signal in the corresponding period's adjustment state, sets the power supplementation reference value, determines the period's supplementation state, records the corresponding supplementation level identifier, and obtains the power supplementation signal level value. The state sequence construction submodule reads the periodic load fluctuation identification signal and the energy storage response trend record based on the power supplementation signal level value, and constructs a state triplet by combining the power supplementation level, load fluctuation amplitude and response trend. After sorting each state triplet by period number, a unified state structure list is constructed to establish the operating state sequence. The strategy scheme generation submodule extracts the cycle number and numerical information corresponding to each state triplet based on the operating state sequence, identifies the degree of deviation of the cycle in terms of load fluctuation amplitude and trend value, sets a stable threshold range, marks the active segment of state change, maps and organizes the data according to the cycle, and obtains the flywheel energy storage management and control scheme.

Images