A data center flywheel energy storage optimization control system

Abstract

The present application relates to the technical field of smart grid, specifically to a data center flywheel energy storage optimization control system.In the present application, the frequency transition characteristics are extracted through real-time identification of the energy consumption curve spectrum distribution, the dynamic response correlation between the rotational speed and the voltage is established at the initial stage of energy storage release, the nonlinear threshold adjustment of the energy storage release path and the fine setting of the rotational speed slope are realized, the asymmetric index is formed by the up-down difference analysis of the periodic power curve, the energy release timing and the response priority are further optimized, the coupling linkage mechanism of the voltage regulation parameter and the energy storage release path is formed, the flywheel can output electric energy in a way that is more consistent with the load change rhythm during the acceleration stage, and finally the fine dynamic regulation of the multi-stage rotational speed slope and the trigger delay of the output process is achieved, which significantly improves the matching degree of power supply stability and energy response, and enhances the adaptability of the system to load fluctuation and power supply switching.

Description

Technical Field

[0001] This invention relates to the field of smart grid technology, and in particular to a data center flywheel energy storage optimization control system. Background Technology

[0002] The field of smart grid technology involves a technical system that utilizes modern information and communication technologies to comprehensively perceive, dynamically optimize, and efficiently control power systems. Its core aspects include real-time monitoring of power systems, data acquisition and processing, equipment operating status assessment, power quality management, and distribution network dispatching and load forecasting.

[0003] Among them, the traditional data center flywheel energy storage optimization control system refers to the operation control mode of flywheel energy storage devices applied to data centers. The technical issues addressed by this system are mainly the large fluctuations in data center power load and poor power supply stability during power switching.

[0004] Traditional flywheel energy storage control systems lack in-depth identification and analysis of energy consumption spectrum changes, making it difficult to identify abrupt changes in load characteristics during the initial startup phase. This results in coarse speed response and output voltage settings, leading to lag in response to sudden increases or decreases in load. In particular, voltage compensation is not timely during power supply switching, easily causing severe voltage fluctuations. Furthermore, these systems fail to perform structured analysis of the up-and-down differences in the power curve, making it impossible to effectively identify the asymmetric characteristics of load fluctuations. Consequently, energy output paths lack priority differentiation during scheduling, and optimal path selection cannot be achieved in multi-path output scenarios. For example, during peak electricity consumption periods or when switching to backup power mode, the voltage output rhythm is out of sync with the load rhythm, causing slow flywheel energy storage response and amplified voltage drops, affecting power supply continuity and system stability. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a data center flywheel energy storage optimization control system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A data center flywheel energy storage optimization control system includes:

[0007] The spectrum transition identification module obtains the real-time energy consumption curve of the driving torque through the energy storage input interface, identifies the transition points of the spectrum amplitude distribution density of the energy consumption curve, and extracts the frequency transition feature parameter set.

[0008] The path adjustment preset module makes a nonlinear preset adjustment to the adjustment threshold of the flywheel energy storage release path with reference to the frequency point transition characteristic parameter set, and sets the speed slope adjustment command.

[0009] The power fluctuation analysis module obtains the power fluctuation curve for a specified period through the energy storage load interface, and compares the asymmetric distribution index of the upward and downward power difference in the power fluctuation curve based on the half-cycle division method to obtain the power asymmetry index.

[0010] The response path filtering module combines the response priority of the speed slope adjustment command and the power asymmetry index with the flywheel energy release path into an inertial intervention activation command.

[0011] The dynamic control execution module dynamically controls the power output state during the flywheel energy storage acceleration phase through the inertial intervention activation command, thereby obtaining the optimized control result of flywheel energy storage.

[0012] As a further aspect of the present invention, the frequency transition characteristic parameter set includes frequency change range, amplitude abrupt change point, and spectral density aggregation position; the speed slope adjustment command includes speed change rate, voltage adjustment threshold, and response timing parameters; the power asymmetry index includes uplink power difference, downlink power difference, and difference distribution difference; the inertial intervention activation command includes adjustment path parameters, voltage threshold value, and output timing setting; and the flywheel energy storage optimization control result includes output power curve, acceleration control state, and voltage adjustment state.

[0013] As a further aspect of the present invention, the spectrum transition identification module includes:

[0014] The energy consumption acquisition submodule obtains the real-time energy consumption curve corresponding to the drive torque through the energy storage input interface, and collects the energy consumption value and time information of each sampling point to construct an energy consumption time-series amplitude data sequence.

[0015] The spectrum conversion submodule obtains the spectrum sequence composed of frequency and amplitude in the energy consumption time-series amplitude data sequence through Fourier transform, performs first-order and second-order difference processing on the amplitude data according to frequency order, and constructs frequency amplitude gradient change curves.

[0016] The inflection point extraction submodule smooths the frequency amplitude gradient change curve, identifies the location of local gradient change rate clusters, determines the boundary of frequency density abrupt change intervals, extracts the corresponding frequency and amplitude data segments, and generates a set of frequency transition feature parameters.

[0017] As a further aspect of the present invention, the path adjustment preset module includes:

[0018] The frequency window identification submodule extracts the frequency peak interval and corresponding amplitude change segment of the frequency point transition feature parameter set, divides the frequency window according to the position of amplitude fluctuation change in the frequency sequence, and establishes a frequency-time mapping segment.

[0019] The voltage mapping construction submodule, based on the frequency-time mapping segment, obtains the response delay and rise slope corresponding to each frequency window as variables, and generates a voltage regulation parameter combination set according to the parameter change process of multiple frequency segments.

[0020] The instruction set setting submodule calculates the inflection point density of the gradient rate of change in the frequency amplitude gradient change curve based on the voltage adjustment parameter combination set, classifies the distribution of the inflection point density in each frequency window into intervals, matches and configures the voltage parameter combination with the flywheel release path, and generates a speed slope adjustment instruction.

[0021] As a further aspect of the present invention, the power fluctuation analysis module includes:

[0022] The power curve acquisition submodule obtains the power fluctuation curve within a specified period through the energy storage load interface, divides the curve into an upward segment and a downward segment according to the power change trend, and generates a partitioned power sequence data group.

[0023] The difference set construction submodule extracts the power extreme values ​​in the uplink and downlink segments of the partitioned power sequence data group, calculates the change amplitude between the power extreme values ​​in each segment, and generates uplink and downlink power difference set pairs.

[0024] The asymmetric index calculation submodule calculates the variability index of the uplink and downlink power difference set pairs respectively, and establishes an asymmetric measurement structure by combining the difference relationship between the two sets of variability indices to generate power asymmetric index.

[0025] As a further aspect of the present invention, the response path filtering module includes:

[0026] The difference extraction submodule extracts the power asymmetry index corresponding to the uplink and downlink power difference set pairs respectively, compares the values ​​of the two sets of variability indexes, obtains the difference between them as the basis for response parameters, and generates variability difference parameters.

[0027] The priority positioning submodule identifies the response priority label corresponding to each flywheel energy release path in the speed slope adjustment command based on the variability difference parameter, compares the variability difference parameter with the priority label set range, and generates a priority filtering path group.

[0028] The intervention instruction generation submodule extracts the voltage threshold parameters and output timing parameters configured for each path in the priority filtering path group, combines and encapsulates the two types of parameters according to the path order, and generates an inertial intervention activation instruction.

[0029] As a further aspect of the present invention, the dynamic control execution module includes:

[0030] The instruction parsing submodule extracts the voltage threshold parameters and output timing parameters set by the inertial intervention activation instruction, parses the parameter combination structure and divides it into application stages, and establishes a set of execution parameter mapping relationships.

[0031] Based on the execution parameter mapping relationship set, the parameter loading submodule loads the voltage threshold value and the corresponding trigger timing sequentially to the flywheel energy storage control structure, sets the effective range and response order of the parameters in each stage of the acceleration phase, and generates the flywheel output control parameter set.

[0032] The output control submodule continuously adjusts the voltage output state of the flywheel energy storage acceleration stage according to the flywheel output control parameter set, changes the relationship between the speed response slope and trigger delay in each stage of the output process, dynamically corrects the rhythm changes of the energy release process, and generates optimized control results for flywheel energy storage.

[0033] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0034] In this invention, frequency transition features are extracted through real-time identification of the energy consumption curve spectrum distribution. A dynamic response correlation between rotational speed and voltage is established immediately at the initial stage of energy storage release. This enables nonlinear threshold adjustment of the energy storage release path and fine setting of the rotational speed slope. Combined with the analysis of the up-down difference of the periodic power curve, an asymmetric index is formed to further optimize the energy release timing and response priority. This promotes a coupling and linkage mechanism between voltage regulation parameters and the energy storage release path, allowing the flywheel to output electrical energy in a way that better matches the rhythm of load changes during the acceleration phase. Ultimately, this achieves fine dynamic control of the rotational speed slope and trigger delay in the multi-stage output process, significantly improving the matching degree of power supply stability and energy response, enhancing the system's adaptability to load fluctuations and power switching, alleviating instantaneous voltage drops and power jitter, and improving the dynamic compensation effect of flywheel energy storage in data center power supply systems. Attached Figure Description

[0035] Figure 1 This is a system flowchart of the present invention;

[0036] Figure 2 This is a flowchart of the spectrum transition identification module of the present invention;

[0037] Figure 3 This is a flowchart of the path adjustment preset module of the present invention;

[0038] Figure 4 This is a flowchart of the power fluctuation analysis module of the present invention;

[0039] Figure 5 This is a flowchart of the response path filtering module of the present invention;

[0040] Figure 6This is a flowchart of the dynamic control execution module of the present invention. Detailed Implementation

[0041] 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.

[0042] 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.

[0043] Please see Figure 1 A data center flywheel energy storage optimization control system includes:

[0044] The spectrum transition identification module obtains the real-time energy consumption curve of the driving torque through the energy storage input interface, identifies the transition points of the spectrum amplitude distribution density of the energy consumption curve, and extracts the frequency transition feature parameter set.

[0045] The path adjustment preset module makes a nonlinear preset adjustment to the adjustment threshold of the flywheel energy storage release path with reference to the frequency point transition characteristic parameter set, and sets the speed slope adjustment command.

[0046] The power fluctuation analysis module obtains the power fluctuation curve for a specified period through the energy storage load interface, and compares the asymmetric distribution index of the upward and downward power difference in the power fluctuation curve based on the half-cycle division method to obtain the power asymmetry index.

[0047] The response path filtering module associates the response priority of the flywheel energy release path with the speed slope adjustment command and the power asymmetry index, and combines them into an inertial intervention activation command.

[0048] The dynamic control execution module dynamically controls the power output state during the acceleration phase of flywheel energy storage by using inertial intervention activation commands, thereby obtaining optimized control results for flywheel energy storage.

[0049] The frequency transition characteristic parameter set includes frequency change range, amplitude abrupt change point, and spectral density clustering location; the speed slope adjustment command includes speed change rate, voltage adjustment threshold, and response timing parameters; the power asymmetry index includes uplink power difference, downlink power difference, and difference distribution difference; the inertial intervention activation command includes adjustment path parameters, voltage threshold, and output timing setting; and the flywheel energy storage optimization control results include output power curve, acceleration control status, and voltage adjustment status.

[0050] Please see Figure 2 The spectrum transition identification module includes:

[0051] The energy consumption acquisition submodule obtains the real-time energy consumption curve corresponding to the drive torque through the energy storage input interface, and collects the energy consumption value and time information of each sampling point to construct an energy consumption time-series amplitude data sequence.

[0052] To obtain the real-time energy consumption curve corresponding to the drive torque through the energy storage input interface, the process requires first deploying an energy storage sensor on the device under test and connecting it to the energy consumption acquisition unit. The sampling system records the input current and voltage of the drive motor at different time points at a fixed frequency and calculates the real-time power. Then, it combines the drive torque information with synchronous recording to form a power curve with time as the horizontal axis and instantaneous power as the vertical axis. Subsequently, the energy consumption value in each time period is obtained by multiplying the power curve by the length of each sampling interval and then summing them up. Each sampling obtains a set of "timestamp + energy consumption value" data, constructing a complete energy consumption time-series amplitude data sequence. For example, if the sampling frequency is set to 100Hz, that is, sampling once every 0.01 seconds, and the sampling time is 10 seconds, then a total of 1000 data points are collected. Each point contains the energy consumption value obtained by multiplying the sampling time and instantaneous power by 0.01 seconds. For example, the 100th sampling point corresponds to a time of 1 second, and its energy consumption is 300W × 0.01s = 3Wh, and so on.

[0053] The spectrum conversion submodule obtains the spectrum sequence composed of frequency and amplitude in the energy consumption time series amplitude data sequence through Fourier transform, performs first-order and second-order difference processing on the amplitude data according to frequency order, and constructs frequency amplitude gradient change curves.

[0054] The frequency-amplitude spectrum of energy consumption time-series data is obtained by Fourier transform. This process requires preprocessing the complete time-series energy consumption data, including normalization and removal of the DC component (average value). Then, the time-domain signal is converted to the frequency domain using Fast Fourier Transform (FFT) to obtain the frequency-amplitude spectrum data. For example, with a sampling frequency of 100Hz and a sampling length of 1024 points, the frequency resolution is 100Hz / 1024≈0.097Hz, resulting in frequency pairs within the range of 0Hz to 50Hz. The amplitude sequence is obtained, and then the amplitude data is sorted in ascending order of frequency. First-order difference is performed on the amplitude data, that is, each term is subtracted from the previous term to obtain the amplitude change gradient curve. Then, second-order difference is performed in the same way, that is, the current first-order difference term is subtracted from the previous term to form the rate of change gradient curve. For example, if the amplitude is 10 at 1Hz and 15 at 1.1Hz, then the first-order difference is 15-10=5. If the amplitude is 18 at 1.2Hz, then the next difference is 18-15=3, and the second-order difference is 3-5=-2. In this way, the frequency amplitude gradient change curve is constructed.

[0055] The inflection point extraction submodule smooths the frequency amplitude gradient change curve, identifies the location of local gradient change rate clusters, determines the boundary of frequency density abrupt intervals and extracts the corresponding frequency and amplitude data segments, and generates a set of frequency point transition feature parameters.

[0056] The frequency amplitude gradient change curve is smoothed. This process usually uses the moving average method or Gaussian filtering to process the original rate of change data into a continuous window to reduce the influence of noise. For example, taking a 5-point moving average, the new value of each frequency point is the average of itself and the two points before and after it. Then, based on the smoothed gradient change curve, the location of local gradient drastic change is identified as a potential spectral abrupt change point. For example, if the rate of change of consecutive points changes from positive to negative or the difference in change exceeds a set threshold (such as 1.5 times the mean), it is marked as a candidate inflection point. Then, the boundary of the frequency density abrupt change interval is determined according to the inflection point position, and the original frequency and amplitude data in the interval are extracted to form a set of segments that can be used for subsequent feature recognition. For example, if the interval between 0.5Hz and 1.5Hz is identified as a jump interval, all frequencies and their corresponding amplitudes in the interval are extracted and summarized into a set of frequency point transition feature parameters.

[0057] Please see Figure 3 The path adjustment preset module includes:

[0058] The frequency window identification submodule extracts the frequency peak interval and corresponding amplitude change segment of the frequency transition feature parameter set, divides the frequency window according to the position of amplitude fluctuation change in the frequency sequence, and establishes frequency-time mapping segment.

[0059] To extract the frequency peak range and corresponding amplitude change segments of the frequency transition feature parameter set, it is necessary to identify the frequency jump points and their frequency value ranges from the previous processing results. This is done by traversing the frequency-amplitude data to find the frequency positions where the change value exceeds a set abrupt change threshold as the peak center. Then, the peak range is determined by expanding to both sides to find the boundary points where the amplitude falls back to the background level. Subsequently, all amplitude data within this range are truncated to form amplitude change segments. For example, if the amplitude of a frequency band suddenly increases to 40 at 20Hz while the adjacent frequency is only 15, then the amplitude is truncated to the left and right of 20Hz as the center. Extend the range by finding the point where the amplitude drops below 20; divide the frequency window according to the position of the amplitude fluctuation change in the frequency sequence, that is, divide the entire spectrum sequence into several frequency segments according to the change point, and record its start and end frequencies and corresponding time indices; on this basis, establish a frequency-time mapping segment, which requires tracing the time tag corresponding to the frequency data, that is, mapping the timestamp before the original spectrum conversion to each frequency window. For example, if the time domain interval corresponding to 20Hz is between t=1.1 and 1.4 seconds, then the time segment mapped by this frequency window is 1.1s–1.4s.

[0060] The voltage mapping construction submodule is based on the frequency-time mapping segment. It obtains the response delay and rise slope corresponding to each frequency window as variables, and generates a set of voltage regulation parameter combinations according to the parameter change process of multiple frequency segments.

[0061] Based on the frequency-time mapping segment, the response delay and rise slope corresponding to each frequency window are obtained as variables. The response delay can be calculated from the difference between the start time of the frequency window and the trigger time of the external excitation signal. Assuming that the excitation signal is emitted at t=1.0 seconds and the frequency window starts at t=1.1 seconds, the response delay is 0.1 seconds. The rise slope is estimated by the amplitude change rate in the frequency window, i.e., Δamplitude / Δtime. For example, if the amplitude increases from 15 to 35 in 0.2 seconds, the rise slope is (35-15) / 0.2=100 amplitude units per second. According to the parameter change process of multiple frequency segments, the response delay and rise slope of different frequency windows are combined in sequence to form a time-progression model. For example, the combination of frequency segments A and B has corresponding response delays of 0.1s and 0.15s, and slopes of 80 and 95, respectively. The combined parameter sequence is {0.1, 80; 0.15, 95}. Finally, a voltage regulation parameter combination set is generated, that is, the above combination sequence is archived in the form of voltage control variables.

[0062] The instruction set setting submodule calculates the inflection point density of the gradient rate of change in the frequency amplitude gradient change curve based on the voltage regulation parameter combination set, classifies the distribution of the inflection point density in each frequency window into intervals, matches and configures the voltage parameter combination with the flywheel release path, and generates the speed slope adjustment instruction.

[0063] Based on the voltage regulation parameter set, the inflection point density of the gradient rate of change in the frequency amplitude gradient curve is calculated. The inflection point density is calculated using the following formula:

[0064] ;

[0065] in, : No. The inflection point density for each frequency window, expressed in "inflections / Hz", represents the number of abrupt changes per unit frequency range within that window. It is obtained by calculating the density of inflection points for each frequency point within the corresponding window in the spectral curve. Determine whether the mutation conditions are met, count the number of mutations, and calculate a weighted average. : Indicates the total number of frequencies included within the current frequency window. Number of frequency sampling points, from point 1 to point 2. Each point is judged and accumulated sequentially. : No. The nth frequency sampling point corresponds to the nth frequency in the spectrum after Fourier transform. Each frequency value, in Hz, is obtained using the following method: ,in This refers to frequency resolution (e.g., a sampling frequency of 100Hz and an FFT point count of 1024). Hz), :frequency The second-order difference at that frequency represents the "acceleration" of the amplitude change at that frequency point, and is calculated as follows: , : No. The amplitude at each frequency point : No. The amplitude at each frequency point : No. The amplitude at each frequency point is derived from the spectral amplitude sequence, obtained from the energy consumption data after FFT processing; the second-order difference unit is "amplitude / Hz". 2 ". : Indicator function, takes a value of 1 if a condition is met, and 0 otherwise. Used to count the number of frequency points that meet the "mutation" condition. The threshold for mutation detection is set for all mutations within the entire frequency band. It is 1.5 times the mean, and this value is adjusted according to the overall intensity of change. It is used to distinguish significant abrupt changes from background fluctuations, and the unit is "amplitude / Hz". 2 ". The frequency weighting function is used to introduce a weighted adjustment of the actual system response rate factor. Its calculation method is as follows: , : No. Each frequency point corresponds to the response rise slope within the frequency window, expressed in "amplitude / s". This represents the rate of energy change within the corresponding frequency range and is obtained by dividing the difference between the maximum amplitude point and the starting point within the frequency window by the corresponding time difference. : Slope adjustment factor, usually an empirical value, such as For example, if the amplitude of a window increases from 15 to 35 in 0.2 seconds, then Corresponding weight . : Frequency window span, in Hz, represents the frequency coverage range of the current window. For example, if the frequency window is from 20Hz to 30Hz, then... Hz. Based on numerous actual measurements, the inflection point density in the typical frequency band of the system is... The following areas will be graded: Low-density zones: The number of pixels per Hz indicates that the frequency window has sparse inflection points and small fluctuations, representing a medium-density range. The value per Hz indicates a certain amplitude abrupt change but overall stability, representing a high-density range. A value per Hz indicates that the spectrum is highly unstable in this frequency range, with significant disturbances, requiring focused control. After calculating the inflection point density, the voltage regulation parameter combination and flywheel release path are matched and configured. Parameters are filtered according to matching rules; for example, in the medium density window, the response delay in the voltage parameter combination is required. seconds, and slope If the amplitude is given in seconds, then records that meet this condition are selected from the set of combinations, for example, combinations. Retained The selected combinations of parameters are then eliminated, and the final combination of parameters that meets the screening criteria is converted into the speed slope adjustment command required by the system control layer.

[0066] Please see Figure 4 The power fluctuation analysis module includes:

[0067] The power curve acquisition submodule obtains the power fluctuation curve within a specified period through the energy storage load interface, divides the curve into an upward segment and a downward segment according to the power change trend, and generates a partitioned power sequence data group.

[0068] To obtain the power fluctuation curve within a specified period through the energy storage load interface, a power sensing device needs to be connected to the target energy storage device, and the sampling period and time interval need to be set. For example, if the sampling is set to 100 times per second and the total period is 10 seconds, a total of 1000 power data points will be collected. The data is acquired by the sensor in real time and uploaded to the data processing unit. The processing unit performs trend analysis on the collected power sequence to determine its overall fluctuation trend. Moving average and first-order difference are used to determine the power change trend. For example, if multiple consecutive sampling points show an upward trend, it is divided into an upward segment, and vice versa. The division process automatically detects inflection points and delineates the boundaries of each segment through a sliding window. The data of each segment includes all power sampling points within that time period. Finally, the complete power curve is split into multiple upward and downward segments, and the segment number and time range are labeled respectively to generate a logically divided partitioned power sequence data group.

[0069] The difference set construction submodule extracts the power extreme values ​​in the uplink and downlink segments of the partitioned power sequence data group, calculates the change amplitude between the power extreme values ​​in each segment, and generates uplink and downlink power difference set pairs.

[0070] Power extrema are extracted from the uplink and downlink segments of the partitioned power sequence data set. Within each uplink segment, the power values ​​corresponding to the starting and ending points are extracted, serving as the minimum and maximum power for that segment, respectively. In the downlink segment, the ending power is used as the minimum value, and the starting point as the maximum value, thus forming a power extrema set for each segment. Next, the variation range between the power extrema within each segment is calculated, i.e., the increment from the starting power to the ending power in the uplink segment, and the decrement from the starting power to the ending power in the downlink segment. For example, if the starting power of an uplink segment is 120W and the ending power is 180W, the variation range is 60W. Similarly, if the starting power of a downlink segment is 190W and the ending power is 140W, the variation range is 50W. Finally, the variation ranges of all uplink and downlink segments are assigned to uplink and downlink power difference sets, forming paired data sets. These sets are then paired according to their sequence positions to generate uplink and downlink power difference set pairs.

[0071] The asymmetric index calculation submodule calculates the variability index of the uplink and downlink power difference sets respectively, and establishes an asymmetric measurement structure by combining the difference relationship between the two sets of variability indices to generate power asymmetric index.

[0072] A variability index is constructed for each set of uplink and downlink power differences, and an asymmetry measurement structure is established based on the difference between the two, thereby generating a power asymmetry index. The core of this module is to quantify the consistency of power fluctuations during uplink and downlink processes, reflect the stability of power differences in each segment through variability, and measure the asymmetry of fluctuations by the relative difference between the two.

[0073] First, the variability index measures the degree of deviation of the differences in each segment of a power difference set from the set average. Essentially, it quantifies dispersion, reflecting the stability of power changes in a specific direction (upward or downward). Its calculation formula is:

[0074] ;

[0075] in, The variability index is a dimensionless value used to describe the intensity of fluctuations within a set. The number of segments contained in the current power difference set is usually determined by dividing the power curve. : No. The power difference value comes from the power extreme value difference extracted in the previous module. For example, if the starting power of a segment is 110W and the ending power is 150W, then the difference is 40W. The average power difference in the set is equal to the power difference of all... The arithmetic mean, : This is an improved stabilization parameter used to prevent the average value from being too high. Approaching zero leads to unstable results; therefore, it is typically set to 4–5 times the power resolution of the sampling device. For example, if the sampling accuracy is 5W, then... W.

[0076] Introduction This allows the variability index to remain stable when dealing with small fluctuations, making it suitable for energy storage scenarios with relatively minor power fluctuations in practical applications. The variability calculation is performed separately on the uplink and downlink difference sets, yielding two index values: uplink variability. with downward variability .

[0077] Based on this, we define an asymmetric index. This is used to measure the relative difference in power fluctuation stability between uplink and downlink directions, and its expression is as follows:

[0078] ;

[0079] in, Power asymmetry index, a dimensionless value; the larger the value, the more inconsistent the fluctuation behavior. , : Represents the degree of fluctuation variation in the upward and downward directions, respectively.

[0080] The sampling period is set to 10 seconds, divided into 5 uplink segments and 5 downlink segments. The power difference unit is watts (W), and the sampling resolution is 5W. W.

[0081] The set of upward interpolation values ​​is: The average value is Substituting into the formula, we get:

[0082] ;

[0083] The set of downlink differences is: The average value is Substituting into the formula, we get:

[0084] ;

[0085] The asymmetric index is: .

[0086] Asymmetric Indicators The range of values ​​is The closer the value is to 0, the more symmetrical it is; the closer it is to 1, the more asymmetrical it is. The classification is based on the following two points: 1. Distribution of experimental data: Analysis of a large amount of measured data from various systems revealed that power fluctuations in most energy storage systems exhibit a certain degree of symmetry during normal operation, with asymmetry indices mostly concentrated between 0.1 and 0.3. Therefore, 0.2 was set as the basic dividing line. 2. System regulation and response capability: The control system has limited sensitivity to fluctuation asymmetry. When the asymmetry index exceeds 0.5, the system's regulation accuracy and response speed decrease significantly. Therefore, 0.5 was used as the boundary for high asymmetry.

[0087] Therefore, the following grading standards are established:

[0088] Symmetric intervals ( The upward and downward fluctuations are extremely close, and the system requires no special adjustments.

[0089] Medium asymmetric intervals ( The fluctuation behavior varies, requiring adjustment of some parameters;

[0090] Highly asymmetric intervals ( The system needs to activate a compensation mechanism or change the control path.

[0091] The asymmetry index is 0.339, falling into the medium asymmetry range, indicating a significant difference between upward and downward fluctuations. A relaxed threshold should be used for the upward path, while stricter response control conditions should be applied for the downward path. Ultimately, the asymmetry index serves as an input to the system's control configuration, participating in the selection logic for the flywheel release strategy or dynamic voltage adjustment path, ensuring the adaptability and stability of the control operation.

[0092] Please see Figure 5 The response path filtering module includes:

[0093] The difference extraction submodule extracts the power asymmetry index corresponding to the uplink and downlink power difference set pairs respectively, compares the values ​​of the two sets of variability indexes, obtains the difference between them as the basis for response parameters, and generates variability difference parameters.

[0094] The module extracts the corresponding power asymmetry indices for each pair of uplink and downlink power difference sets, using them as core reference items. It then calls upon the calculated uplink and downlink variability indices and compares their values ​​item by item. Specifically, this module performs index matching on the asymmetry indices extracted from each pair of uplink and downlink power difference sets, reads the uplink and downlink variability values ​​from that pair, and directly performs the difference operation. For example, if the uplink variability is 0.0021 and the downlink variability is 0.0012, the variability difference is 0.0009. This value is input into the path response dynamic evaluation parameter control module, which records the path number and the difference value pair, and outputs the variability difference parameter set for all paths in a list format.

[0095] The priority positioning submodule identifies the response priority label corresponding to each flywheel energy release path in the speed slope adjustment command based on the variability difference parameter, compares the variability difference parameter with the priority label set range, and generates a priority filtering path group.

[0096] Based on the variability difference parameter, the path response level identification in the speed slope adjustment command is executed. This process assigns different numerical ranges of variability difference to response priorities by setting predefined priority label ranges. For example, a difference below 0.001 is "high priority," between 0.001 and 0.003 is "medium priority," and above 0.003 is "low priority." The system sequentially reads the variability difference parameter of each flywheel energy release path, compares it with the above range, matches the corresponding priority label, and appends the label to the original path identifier. Simultaneously, when multiple paths have the same priority label, a secondary sort is performed by combining the path number and response time information, ultimately generating a priority-filtered path group, i.e., filtering out all flywheel energy release paths belonging to the same priority level.

[0097] The intervention instruction generation submodule extracts the voltage threshold parameters and output timing parameters configured for each path in the priority filtering path group, combines and encapsulates the two types of parameters according to the path order, and generates an inertial intervention activation instruction.

[0098] The system extracts the voltage threshold parameters and output timing parameters configured for each path in the priority-selected path group, and reads and structures these two types of parameters according to the path arrangement order. The voltage threshold parameters include the activation voltage threshold and the corresponding upper limit of the change response voltage for each path; the lower limit is determined according to the energy storage system's setting standards. The timing parameters include indicators such as the action delay, duration, and recovery time for each path. During the encapsulation process, the system constructs a complete path parameter body using the path number as the primary key, embedding the voltage threshold and timing parameters into a unified structure, and arranging them in descending order of priority. Finally, an inertial intervention activation instruction set is generated, with each instruction containing the path number, voltage control range, and activation timing control parameters.

[0099] Please see Figure 6 The dynamic control execution module includes:

[0100] The instruction parsing submodule extracts the voltage threshold parameters and output timing parameters set by the inertial intervention activation instruction, parses the parameter combination structure and divides it into application stages, and establishes a set of execution parameter mapping relationships.

[0101] The system extracts the voltage threshold parameters and output timing parameters set in the inertial intervention activation command. For each command, fields such as path number, voltage range (including start and end thresholds), and response time period (including delayed start time and duration) are read and categorized. After parsing, the commands are divided into different application stages according to their objectives, such as the pre-start stage, acceleration stage, and stability maintenance stage. The system assigns labels to each set of parameters based on stage characteristics and constructs an execution mapping relationship in the control task scheduling table. For example, if path A has a voltage threshold of 450V–490V and a timing of 0.3s delay + 1.2s duration, the system maps this set of parameters to the initial stage of the acceleration phase, sets its priority and adjustment range, and ultimately establishes a unified set of execution parameter mapping relationships.

[0102] The parameter loading submodule loads the voltage threshold value and the corresponding trigger timing sequentially to the flywheel energy storage control structure based on the execution parameter mapping relationship set, sets the effective range and response order of the parameters in each stage of the acceleration phase, and generates the flywheel output control parameter set.

[0103] Based on the parameter mapping set, voltage threshold values ​​and corresponding triggering sequences are sequentially loaded into the flywheel energy storage control structure. The system loads voltage control information and response times for different paths into the control unit according to the application stage and priority of each parameter. The system also sets the effective range for each parameter, clarifying the stage at which the voltage signal takes effect, the end time, and the response condition judgment logic. A parameter trigger matrix is ​​established in the control structure, and the acceleration stage is segmented. For example, the initial control threshold is 440V–460V with a corresponding response delay of 0.2 seconds; the middle stage is 470V–500V with a corresponding duration of 1.5 seconds; and the final stage is 480V–495V with an effective period of 0.6 seconds. All parameters are registered by a unified scheduling program to form a flywheel output control parameter set. Each record includes its effective range, response delay, and corresponding path number, ensuring that the control logic maintains consistency with the parameter loading order and the control layer execution.

[0104] The output control submodule continuously adjusts the voltage output state of the flywheel energy storage acceleration stage according to the flywheel output control parameter set, changes the relationship between the speed response slope and trigger delay at each stage of the output process, dynamically corrects the rhythm changes of the energy release process, and generates optimized control results for flywheel energy storage.

[0105] Based on the flywheel output control parameter set, the voltage output state during the flywheel energy storage acceleration phase is continuously adjusted. The module controls the flywheel output response behavior segment by segment according to the loaded parameter range and timing settings. In actual operation, the system monitors the current voltage state and time nodes in real time to determine whether the parameter triggering conditions are met. If the triggering is successful, the flywheel output speed adjustment range is controlled to change the response slope of that phase. For example, the initial slope is set to 50 rpm / s and the voltage change rate is 20V / s. After entering the next phase, it is adjusted to 70 rpm / s and 35V / s. The response delay of each phase is dynamically corrected. For example, if the system delay deviates from the preset value by more than 5%, the starting point of the next phase will be adjusted. The system as a whole establishes a rhythm control closed loop, dynamically optimizing the power release mode and speed response structure during the flywheel output process. After the output of each phase is completed, all adjustment behaviors are summarized to generate a flywheel energy storage optimization control result with stronger power response coordination.

[0106] 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 data center flywheel energy storage optimization control system, characterized in that, The system includes: The spectrum transition identification module obtains the real-time energy consumption curve of the driving torque through the energy storage input interface, identifies the transition points of the spectrum amplitude distribution density of the real-time energy consumption curve, and extracts the frequency point transition feature parameter set. The path adjustment preset module makes a nonlinear preset adjustment to the adjustment threshold of the flywheel energy storage release path with reference to the frequency point transition characteristic parameter set, and sets the speed slope adjustment command. The power fluctuation analysis module obtains the power fluctuation curve for a specified period through the energy storage load interface, and compares the asymmetric distribution index of the upward and downward power difference in the power fluctuation curve based on the half-cycle division method to obtain the power asymmetry index. The response path filtering module combines the response priority of the speed slope adjustment command and the power asymmetry index with the flywheel energy release path into an inertial intervention activation command. The dynamic control execution module dynamically controls the power output state during the flywheel energy storage acceleration phase through the inertial intervention activation command, thereby obtaining the optimized control result of flywheel energy storage. The frequency transition characteristic parameter set includes frequency change range, amplitude abrupt change point, and spectral density aggregation position; the speed slope adjustment command includes speed change rate, voltage adjustment threshold, and response timing parameters; the power asymmetry index includes uplink power difference, downlink power difference, and difference distribution difference; the inertial intervention activation command includes adjustment path parameters, voltage threshold value, and output timing setting; and the flywheel energy storage optimization control result includes output power curve, acceleration control state, and voltage adjustment state. The spectral transition identification module includes: The energy consumption acquisition submodule obtains the real-time energy consumption curve corresponding to the drive torque through the energy storage input interface, and collects the energy consumption value and time information of each sampling point to construct an energy consumption time series amplitude data sequence. The spectrum conversion submodule obtains the spectrum sequence composed of frequency and amplitude in the energy consumption time-series amplitude data sequence through Fourier transform, performs first-order and second-order difference processing on the amplitude data according to frequency order, and constructs frequency amplitude gradient change curves. The inflection point extraction submodule smooths the frequency amplitude gradient change curve, identifies the location of local gradient change rate clusters, determines the boundary of frequency density abrupt change intervals, extracts the corresponding frequency and amplitude data segments, and generates a set of frequency transition feature parameters.

2. The data center flywheel energy storage optimization control system according to claim 1, characterized in that, The path adjustment preset module includes: The frequency window identification submodule extracts the frequency peak interval and corresponding amplitude change segment of the frequency point transition feature parameter set, divides the frequency window according to the position of amplitude fluctuation change in the frequency sequence, and establishes a frequency-time mapping segment. The voltage mapping construction submodule, based on the frequency-time mapping segment, obtains the response delay and rise slope corresponding to each frequency window as variables, and generates a voltage regulation parameter combination set according to the parameter change process of multiple frequency segments. The instruction set setting submodule calculates the inflection point density of the gradient rate of change in the frequency amplitude gradient change curve based on the voltage adjustment parameter combination set, classifies the distribution of the inflection point density in each frequency window into intervals, matches and configures the voltage parameter combination with the flywheel release path, and generates a speed slope adjustment instruction.

3. The data center flywheel energy storage optimization control system according to claim 1, characterized in that, The power fluctuation analysis module includes: The power curve acquisition submodule obtains the power fluctuation curve within a specified period through the energy storage load interface, divides the curve into an upward segment and a downward segment according to the power change trend, and generates a partitioned power sequence data group. The difference set construction submodule extracts the power extreme values ​​in the uplink and downlink segments of the partitioned power sequence data group, calculates the change amplitude between the power extreme values ​​in each segment, and generates uplink and downlink power difference set pairs. The asymmetric index calculation submodule calculates the variability index of the uplink and downlink power difference set pairs respectively, and establishes an asymmetric measurement structure by combining the difference relationship between the two sets of variability indices to generate power asymmetric index.

4. The data center flywheel energy storage optimization control system according to claim 3, characterized in that, The response path filtering module includes: The difference extraction submodule extracts the power asymmetry index corresponding to the uplink and downlink power difference set pairs respectively, compares the values ​​of the two sets of variability indexes, obtains the difference between the two sets of variability indexes as the basis for response parameters, and generates variability difference parameters. The priority positioning submodule identifies the response priority label corresponding to each flywheel energy release path in the speed slope adjustment command based on the variability difference parameter, compares the variability difference parameter with the priority label set range, and generates a priority filtering path group. The intervention instruction generation submodule extracts the voltage threshold parameters and output timing parameters configured for each path in the priority filtering path group, combines and encapsulates the two types of parameters according to the path order, and generates an inertial intervention activation instruction.

5. The data center flywheel energy storage optimization control system according to claim 4, characterized in that, The dynamic control execution module includes: The instruction parsing submodule extracts the voltage threshold parameters and output timing parameters set by the inertial intervention activation instruction, parses the parameter combination structure and divides it into application stages, and establishes a set of execution parameter mapping relationships. Based on the execution parameter mapping relationship set, the parameter loading submodule loads the voltage threshold value and the corresponding trigger timing sequentially to the flywheel energy storage control structure, sets the effective range and response order of the parameters in each stage of the acceleration phase, and generates the flywheel output control parameter set. The output control submodule continuously adjusts the voltage output state of the flywheel energy storage acceleration stage according to the flywheel output control parameter set, changes the relationship between the speed response slope and trigger delay in each stage of the output process, dynamically corrects the rhythm changes of the energy release process, and generates optimized control results for flywheel energy storage.

Images