Flywheel and battery hybrid energy storage system multimode cross data fusion scheduling method

Abstract

The invention provides a multi-mode cross data fusion scheduling method for a flywheel and battery hybrid energy storage system, and the method comprises the steps: collecting multi-mode signals in parallel, and fusing and packaging the multi-mode signals into a multi-mode data vector with a unified time sequence. Preprocessing and normalizing the multi-mode data vector to obtain a preprocessing sequence, and extracting a multi-dimensional grid-connected feature from the preprocessing sequence; and constructing a feature association graph based on the multi-dimensional grid-connected features, and fusing node risks of the feature association graph through a linear diffusion model to determine a grid-connected readiness score. And mapping the grid connection preparation degree score to a preset interval to determine a slope starting time window. And performing slope parameterization processing based on the slope starting time window to construct a power step sequence, and generating a plurality of control packets for issuing instructions according to power steps according to the power step sequence. According to the method, by generating a progressive grid-connected power slope, stage injection of short-circuit current is achieved, and one-time large current impact is avoided.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of distributed energy and micro-grid technology, and more particularly, relates to a multi-mode cross data fusion scheduling method for a flywheel and battery hybrid energy storage system. BACKGROUND

[0002] With the rapid development of distributed energy and micro-grid technology, off-grid micro-grid plays an increasingly important role in scenarios such as grid maintenance, remote area power supply, and emergency rescue. In order to better smooth the switching between micro-grid and public grid, autonomous energy storage system becomes a key support. Flywheel energy storage can instantly absorb or release large power impact due to its millisecond-level power response, super-long cycle life and high-frequency energy release characteristics; lithium battery takes the task of continuous energy balance with high energy density and minute-level endurance. The mixed network of the two can be "fast-slow" complementary: flywheel responds to transient impact of grid connection, battery completes subsequent steady-state compensation, and together guarantees the smooth transition of voltage, frequency and power of micro-grid in the process of switching and grid connection. The multi-mode cross data fusion scheduling method further realizes the gradual power ramp injection and absorption through the joint evaluation of grid frequency, bus phase, flywheel power, SOC and thermal management state, and provides intelligent protection for micro-grid grid connection.

[0003] Currently, in the switching and grid connection moment of micro-grid island and main grid, the existing technology includes: (1) Fixed dead zone soft start: by setting a fixed power dead zone for grid connection, the current impact of the moment of closing is reduced, but since the dead zone parameter is difficult to adjust online once it is set, it cannot take into account the impact amplitude difference caused by different bus impedance and load changes; (2) Static reactive power / active power compensation device: STATCOM, SVC and other devices can provide continuous reactive power support and partial active power balance, but their response speed is mostly in tens to hundreds of milliseconds, which is difficult to cover the millisecond-level over-limit short-circuit pulse in micro-grid grid connection; (3) Single energy storage unit isolated buffer: pure flywheel or pure battery system respectively undertakes impact or endurance, but a single system is difficult to meet the high-rate short-time and long-period energy demand at the same time, and lacks multi-source data fusion scheduling capability, and cannot dynamically segment manage the pulse amplitude and duration.

[0004] From the above prior art, it can be seen that the existing scheme either responds too slowly, or the energy capacity is insufficient, or it lacks segmented scheduling logic for grid-connected transient impact, and it is difficult to balance the pulse suppression effect and energy storage unit health maintenance. The data acquisition and fusion evaluation technology of multi-mode cross is still in the exploratory stage, and there is a lack of mature solutions with high maturity, low integration difficulty and adaptability to the variable grid-connected environment of microgrid. Therefore, at the moment of microgrid island switching and grid connection, due to the phase and frequency deviation between the grid and the bus, the short-circuit current peak generated by the closing of the bus in an instant far exceeds the rated capacity. If not through the multi-mode cross fusion scheduling of flywheel and battery, the phased gradual power injection and absorption, the over-limit pulse will tear the bus fuse and trigger the protection trip, resulting in the loss of power of the entire microgrid. SUMMARY

[0005] In order to solve the problems in the prior art, the purpose of the present application is to solve the above-mentioned defects, and further to provide a multi-mode cross data fusion scheduling method for a flywheel and battery hybrid energy storage system.

[0006] The application adopts the following technical solutions.

[0007] The first aspect of the application discloses a multi-mode cross data fusion scheduling method for a flywheel and battery hybrid energy storage system, the method comprising: parallelly acquiring multi-mode signals, and fusing and packaging the multi-mode signals into a multi-mode data vector with unified time sequence; preprocessing and normalizing the multi-mode data vector to obtain a preprocessing sequence, and extracting multi-dimensional grid-connected features from the preprocessing sequence; constructing a feature correlation graph based on the multi-dimensional grid-connected features, and fusing the node risks of the feature correlation graph through a linear diffusion model to determine a grid-connected preparation degree score; mapping the grid-connected preparation degree score to a preset interval to determine a ramp start time window; performing ramp parameterization processing based on the ramp start time window to construct a power ladder sequence, and generating a plurality of control packets according to the power ladder sequence, wherein the control packets are issued according to the power ladder; The multi-mode signals are bus frequency and phase signals on a public alternating current bus, flywheel output power of each flywheel energy storage converter alternating current side, battery state of charge and maximum discharge power, and cooler output power. The upper limit value of the preset interval is the longest ramp start delay time, and the lower limit value is the shortest ramp start delay time.

[0008] Further, the parallel acquisition of multi-mode signals and the fusion and packaging of the multi-mode signals into a multi-mode data vector with unified time sequence comprise: A frequency transmitter and a synchronous phase measurement unit are arranged on the public communication bus, and a synchronous sampling method is adopted to collect the bus frequency and phase signals with time stamps in combination with a GPS clock; A voltage sensor and a current sensor are arranged on the AC side of each flywheel energy storage converter, and a power calculation formula is executed by a power calculation unit to collect the flywheel output power with time stamps.

[0009] Further, the parallel collection of multi-mode signals and the fusion and packaging of the multi-mode signals into a unified time sequence multi-mode data vector also include: The SOC is updated in real time in the battery management unit by Coulomb counting, and the maximum discharge power of the battery is estimated based on the battery temperature and internal resistance lookup table, and is issued by CAN or EtherCAT bus to obtain the battery state of charge and maximum discharge power; Flow meters and temperature difference sensors are arranged in the cooling circuits of the flywheel cabin and the battery cabinet to calculate the cooling power in real time, and the cooling power is used to determine the output power of the cooler; The multi-mode signals are clock-synchronized by the host computer, resampled and interpolated to align at a preset frequency, fused and packaged to obtain the multi-mode data vector.

[0010] Further, the multi-mode data vector is preprocessed and normalized to obtain a preprocessing sequence, and multi-dimensional grid-connected features are extracted from the preprocessing sequence, including: Each component in the multi-mode data vector is subjected to mutation threshold testing in the central control unit to mark and remove abnormal data, and linear interpolation is used to fill in missing values; The multi-mode data vector after the abnormal data and missing value filling is subjected to noise filtering and smoothing processing, and is resampled using a unified time grid to normalize the multi-mode data vector to obtain a normalized preprocessing sequence.

[0011] Further, the multi-mode data vector is preprocessed and normalized to obtain a preprocessing sequence, and multi-dimensional grid-connected features are extracted from the preprocessing sequence, including: The bus frequency sequence is extracted from the preprocessing sequence, and the frequency change rate is calculated at a fixed time interval in the real-time processor to obtain a frequency change rate sequence; The phase signal sequence is extracted from the preprocessing sequence, and the absolute value of the phase difference of each sampling point in the phase signal sequence is calculated to construct an absolute phase deviation sequence.

[0012] Further, the multi-mode data vector is preprocessed and normalized to obtain a preprocessing sequence, and multi-dimensional grid-connected features are extracted from the preprocessing sequence, including: extracting a flywheel output power sequence from the pre-processed sequence and calculating a flywheel output power increment at the same time interval through a controller; extracting a battery state of charge and maximum discharge power sequence from the pre-processed sequence and fusing the battery state of charge and maximum discharge power in a BMS to obtain a battery remaining available power indicator; extracting a cooler output power sequence from the pre-processed sequence and calculating a cooling margin coefficient according to the rated refrigeration power of the cooling unit.

[0013] Further, the multi-dimensional grid connection feature is constructed based on the feature correlation graph, and the node risk of the feature correlation graph is fused through a linear diffusion model to determine the grid connection readiness score, comprising: Based on the multi-dimensional grid connection feature, a multi-dimensional feature vector is constructed, and each dimension feature is disturbed in a microgrid digital simulation platform to determine the weight coefficient corresponding to each dimension feature; According to the set threshold value of each dimension feature, the initial risk of each dimension feature is calculated to construct an initial risk vector, and the feature correlation graph is constructed based on the weight coefficient and the initial risk vector.

[0014] Further, the multi-dimensional grid connection feature is constructed based on the feature correlation graph, and the node risk of the feature correlation graph is fused through a linear diffusion model to determine the grid connection readiness score, further comprising: The linear diffusion model is called to iteratively update the initial risk of each dimension feature, and the fused node risk is output after convergence; Based on the fused node risk corresponding to each dimension feature, an overall risk index is calculated, and the grid connection readiness score is calculated according to the overall risk index; The overall risk index is the weighted average of the fused node risk corresponding to each dimension feature.

[0015] Further, the slope parameterization processing is performed based on the slope start time window to construct a power step sequence, and a plurality of control packets according to the power step sequence are generated, comprising: obtaining a bus rated power capacity, calculating a rated power peak upper limit based on the grid connection readiness score and the bus rated power capacity, the rated power peak upper limit being the product of the grid connection readiness score and the bus rated power capacity; setting the proportion of the primary grid injection power to the rated power peak upper limit as the primary slope amplitude ratio, and calculating the total number of slopes combined with the sampling period to perform slope parameterization processing.

[0016] Further, the slope parameterization processing based on the ramp start time window is performed to construct a power step sequence, and a plurality of control packets according to the power step sequence are generated according to the power step sequence, and the method further comprises: calculating a primary power based on the primary ramp amplitude ratio and the upper limit of the rated power peak value, and calculating a primary time length and a primary step number according to the ramp start time window; generating a step based on the power increment of each step in the remaining ramp step number other than the primary step number, and constructing the power step sequence according to the generated step; wherein each power step in the power step sequence corresponds to a control packet of a step-down instruction.

[0017] The second aspect of the application discloses a flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling system, which is used to realize the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method of any one of the first aspect, and the system comprises: a data acquisition and fusion module, which is used to acquire multi-mode signals in parallel, and fuse and package the multi-mode signals into a unified time sequence multi-mode data vector; a feature extraction module, which is used to pre-process and normalize the multi-mode data vector to obtain a pre-processing sequence, and extract multi-dimensional grid-connected features from the pre-processing sequence; a grid-connected readiness score module, which is used to construct a feature correlation graph based on the multi-dimensional grid-connected features, and fuse the node risk of the feature correlation graph through a linear diffusion model to determine a grid-connected readiness score; a ramp window determination module, which is used to map the grid-connected readiness score to a preset interval to determine a ramp start time window; an instruction issuing module, which is used to perform slope parameterization processing based on the ramp start time window to construct a power step sequence, and generate a plurality of control packets according to the power step sequence according to the power step sequence; wherein the multi-mode signals are bus frequency and phase signals on a public alternating current bus, flywheel output power on each flywheel energy storage converter alternating current side, battery state of charge and maximum discharge power, and cooler output power; the upper limit of the preset interval is the longest ramp start delay time, and the lower limit is the shortest ramp start delay time.

[0018] The third aspect of the application discloses a terminal, comprising a processor and a storage medium; the storage medium is used to store instructions; The processor is used to operate according to the instructions to perform the steps of the method of the first aspect.

[0019] The fourth aspect of the present application discloses a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the steps of the method of the first aspect.

[0020] The present application has the following advantages compared with the prior art: (1) The present application collects the bus frequency and phase signal on the common AC bus, the flywheel output power of each flywheel energy storage converter AC side, the battery state of charge and maximum discharge power, and the cooler output power to form a multi-mode signal, and then fuses the multi-mode signal into a multi-mode data vector with uniform timing, so as to realize the collection of multiple key signals related to grid-connected transient, and provide a data basis for subsequent multi-dimensional feature extraction and gradual grid-connected power ramp generation.

[0021] (2) The present application pre-processes and normalizes the collected multi-mode data vector, and extracts multi-dimensional grid-connected features from the obtained pre-processed sequence, so as to eliminate the deviation caused by different physical dimensions and value ranges, so that the subsequent direct weighted summation of risk indicators (such as the overall risk index) has comparability and physical meaning on data, and can more conveniently reflect the equal contribution of each dimensional feature to the overall risk under the equal weight assumption.

[0022] (3) The present application constructs a feature correlation graph based on the multi-dimensional grid-connected features, and fuses the node risk of the feature correlation graph through a linear diffusion model to determine the grid-connected preparation degree score, and then determines the ramp start time window, finally performs parameterization processing on the ramp start time window to construct a power ladder sequence, and generates multiple control packets according to the power ladder issuing instructions. In the microgrid island switching grid-connected moment, the designed power ladder control scheme is gradually executed, which together with the grid-connected preparation degree score and the ramp window forms a closed-loop scheduling, through the multi-mode cross fusion scheduling of flywheel and battery, the phased gradual power injection and absorption, the problem of tearing the bus fuse and triggering protection trip caused by the super-limit pulse is solved. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 A flowchart of the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the present application is shown in the figure. Figure 2 A system architecture schematic diagram of the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling provided by the present application is shown in the figure. DETAILED DESCRIPTION

[0024] The present application will be further described below in conjunction with the drawings. The following examples are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0025] AsFigure 1 As shown in one embodiment, a flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method includes the following steps: Step S110, parallel acquisition of multi-mode signals, and fusion and packaging of the multi-mode signals into a unified time sequence multi-mode data vector.

[0026] Among them, the multi-mode signals are bus frequency and phase signals on the public AC bus, flywheel output power on the AC side of each flywheel energy storage converter, battery state of charge and maximum discharge power, and cooler output power.

[0027] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the present application specifically includes the following steps in step S110: Step S111, a frequency transmitter and a synchronous phase measurement unit are arranged on the public AC bus, and a synchronous sampling method is used in combination with a GPS clock to acquire bus frequency and phase signals with time stamps.

[0028] Step S112, voltage sensors and current sensors are arranged on the AC side of each flywheel energy storage converter, a power calculation formula is executed through a power calculation unit, and flywheel output power with time stamps is acquired.

[0029] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the present application specifically further includes the following steps in step S110: Step S113, the SOC is updated in real time in the battery management unit through Coulomb counting, and the maximum discharge power of the battery is estimated based on a battery temperature and internal resistance lookup table, and is issued through a CAN or EtherCAT bus to acquire the battery state of charge and the maximum discharge power.

[0030] Step S114, flow meters and temperature difference sensors are arranged in the cooling circuits of the flywheel cabin and the battery cabinet to calculate the cooling power in real time, and the cooler output power is determined based on the cooling power.

[0031] Step S115, the multi-mode signals are clock-synchronized by the host computer, resampled and interpolated to align at a preset frequency, fused and packaged to obtain a multi-mode data vector.

[0032] In a specific embodiment, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the present application includes steps 1-4: Step 1, parallel acquisition and preprocessing of multi-mode signals.

[0033] including the following steps: Step 1.1, acquisition of bus frequency and phase signals.

[0034] Specifically, a high-precision frequency transmitter and a synchronous phase measurement unit are installed on the public exchange bus, a synchronous sampling technology (sampling rate 1 kHz) is adopted, and an absolute timestamp is given to each measurement through the GPS clock, revealing the frequency difference and phase difference between the grid and the microgrid at the instant of grid connection, providing a reference for subsequent slope grid connection judgment.

[0035] Step 1.2, flywheel output power measurement.

[0036] Specifically, high-precision voltage and current sensors (sampling rate 2 kHz) are installed on each flywheel energy storage converter AC side, a power calculation unit is configured to calculate the flywheel output power according to the formula power=voltage x current, and is time-stamped synchronously. This step quantifies the flywheel's absorption of the bus's electric capacity to determine the available buffer and response speed.

[0037] Step 1.3, battery state of charge and output capacity collection.

[0038] Specifically, the battery management system updates the SOC every second through the coulomb counting method and estimates the maximum discharge power based on temperature and internal resistance lookup table, and issues the battery SOC value and the maximum available discharge power through the vehicle-level CAN or EtherCAT bus to evaluate the battery's sustained output capacity and determine the subsequent battery's time window and power share.

[0039] Step 1.4, thermal management system state collection.

[0040] Specifically, flow meters and temperature difference sensors are arranged in the cooling circuits of the flywheel cabin and the battery cabinet to calculate the cooling power (flow x specific heat capacity x temperature difference) in real time, sample at 100 Hz and return to obtain the cooling power index of the cooler, which is used to judge the thermal margin during high-power pulse to avoid overheating.

[0041] Step 1.5, data fusion and packaging.

[0042] Specifically, the main control machine synchronizes the various dimensions of data collected in steps 1.1~1.4, then resamples and interpolates at a frequency of 100 Hz, and finally forms a 6 x N matrix, where the rows correspond to signal types and the columns correspond to time points, i.e. a unified time sequence multi-modal data vector. This step constructs a high-dimensional, multi-modal raw data set, providing comprehensive and synchronous input for subsequent grid connection impact buffer decision-making.

[0043] The 6 x N matrix formed is The expression is: ; In the formula, is the bus instantaneous frequency value at time t, is the phase value at time t, P(t) is the flywheel output power at time t, SOC(t) is the SOC value at time t, Pmax(t) is the maximum available discharge power of the battery at time t, Pcool(t) is the cooling power index at time t, T represents the transpose of the matrix, and N represents the total number of time stamps.

[0044] Step 1.6, outlier detection and elimination.

[0045] Specifically, the central control unit composed of FPGA and DSP applies the upper and lower limit based "mutation threshold" test to each component in the multi-mode data vector after data fusion and packaging: (1) The bus frequency exceeds 47-53Hz; (2) The phase mutation amplitude exceeds 10° / ms; (3) The flywheel / battery power change rate is greater than 50% / ms; (4) The cooling power mutation exceeds ±30% / s.

[0046] If (1)-(4) are met, the data at the corresponding time is marked as abnormal data and eliminated.

[0047] At the same time, the linear interpolation of the previous or next time of the detected missing value is filled to ensure the continuity of the data.

[0048] Step 1.7, noise filtering and smoothing.

[0049] Specifically, the sliding median filter is used for the DSP in the cabinet, the median filter with a window length of 5 points is used for the frequency and phase channels, and the weighted moving average with a window length of 7 points is used for the power and battery SOC, cooling power channels.

[0050] Step 1.8, uniform time grid resampling.

[0051] Specifically, the sampling period is set to 10ms in the main control server, and the data sampled at the original time stamp which is not an integer multiple of 10ms is interpolated to , , , is the original time stamp. At the same time, the new time stamp after resampling is recorded.

[0052] Step 1.9, amplitude normalization and calibration.

[0053] Specifically, according to the system specifications, the maximum and minimum safe values of each signal are given, and each channel is mapped to the interval [0,1] according to its safe interval, outputting dimensionless signals. All calibration coefficients are saved in the control cabinet EEPROM, supporting online update.

[0054] In the embodiment, the data preprocessing and normalization eliminate the deviation caused by different physical dimensions and value ranges, so that the subsequent direct weighted sum of risk indicators (for example: overall risk index) has mathematical comparability and physical meaning, thereby being able to more conveniently reflect the equivalent contribution of each feature to the overall risk under the equal weight assumption.

[0055] In step S120, the multi-mode data vector is preprocessed and normalized to obtain a preprocessed sequence, and multi-dimensional grid-connected features are extracted from the preprocessed sequence.

[0056] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application comprises the following steps in step S120: In step S121, the central control unit performs mutation threshold test on each component in the multi-mode data vector to mark and remove abnormal data, and performs linear interpolation filling on missing values.

[0057] In step S122, the multi-mode data vector after removing abnormal data and filling missing values is subjected to noise filtering and smoothing processing, and is resampled by using a unified time grid to normalize the multi-mode data vector to obtain a normalized preprocessed sequence.

[0058] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application further comprises the following steps in step S120: In step S123, the bus frequency sequence is extracted from the preprocessed sequence, and the frequency change rate is calculated at a fixed time interval in the real-time processor to obtain a frequency change rate sequence.

[0059] In step S124, the phase signal sequence is extracted from the preprocessed sequence, and the absolute value of the phase difference of each sampling point in the phase signal sequence is calculated to construct an absolute phase difference sequence.

[0060] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application further comprises the following steps in step S120: In step S125, the flywheel output power sequence is extracted from the preprocessed sequence, and the flywheel output power increment is calculated at the same time interval by the controller.

[0061] In step S126, the battery state of charge and maximum discharge power sequence are extracted from the preprocessed sequence, and the battery state of charge and maximum discharge power are fused in the BMS to obtain a battery remaining available power indicator.

[0062] Step S127, extract the cooler output power sequence from the pre-processing sequence, and calculate the cooling margin coefficient according to the rated refrigeration power of the cooling unit.

[0063] In specific embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application comprises the following steps 2: key grid-connected feature extraction.

[0064] Comprising the following steps: Step 2.1, frequency change rate Extraction.

[0065] Specifically, in the real-time processor, the normalized frequency value at the nth time is calculated according to the fixed time interval Calculation: ; In the formula, respectively, are the normalized frequency values at the nth and n-1 time.

[0066] Then, the sampling processor compares with the preset threshold value (such as 0.02 / s), and records the frequency slope moment exceeding the threshold value. This step quantifies the frequency jump speed at the grid-connected moment, and is used to evaluate the response speed of the system to suppress the short-circuit pulse.

[0067] Step 2.2, phase deviation quantification.

[0068] Specifically, the absolute value of the phase difference of each sampling point in the normalized phase difference sequence is calculated to obtain the corresponding non-negative value. In the phase measurement unit, when the non-negative value exceeds the set safe angle (such as 0.1, i.e. 36°), a high deviation flag is triggered, which can directly reflect the degree of phase asynchronization of the microgrid and the main grid voltage waveform, and is used to determine the grid-connected phase tolerance.

[0069] Step 2.3, flywheel power increment calculation.

[0070] Specifically, the flywheel power increment is calculated by the controller according to the same time difference of the normalized flywheel power sequence. If the flywheel power increment is greater than 0.05 (corresponding to 5% rated power / 10ms), it is marked as "large amplitude power jump", which is used to capture the flywheel transient charge and discharge rate, and evaluate the acceptable acceleration and deceleration amplitude in the slope grid connection.

[0071] Step 2.4, battery available power evaluation.

[0072] Specifically, in the BMS, the normalized SOC sequence and the SOC in the battery output power are combined with the nominal maximum output power (constant), i.e. multiplied by each other, to obtain the battery remaining available power index. When the battery remaining available power index is lower than 10% of the nominal maximum output power, the battery capacity is insufficient.

[0073] Step 2.5, cooler heat margin calculation.

[0074] Specifically, taking the rated cooling power of the cooling system as the reference, the cooling margin coefficient is calculated, and the expression is: ; In the formula, , the nth timestamp in the normalized cooler output power sequence corresponds to the cooler output power.

[0075] If is less than 0.2, a warning is given that the thermal management is approaching the limit, which can be used to reflect the remaining cooling capacity of the thermal management system during the grid-connection impact, to ensure that the flywheel and the battery do not decrease in power due to overheating.

[0076] Finally, based on the frequency slope , the phase deviation , the flywheel power jump , the available power of the battery and the cooling margin obtained in steps 2.1-2.5 during the grid-connection transient state, a multi-dimensional grid-connection feature vector is constructed, which is expressed as: ; Step S130, based on the multi-dimensional grid-connection feature, a feature correlation graph is constructed, and the node risk of the feature correlation graph is fused through a linear diffusion model to determine the grid-connection preparation degree score.

[0077] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application specifically includes the following steps: Step S131, based on the multi-dimensional grid-connection feature, a multi-dimensional feature vector is constructed, and each dimension of the feature is disturbed in the microgrid digital simulation platform to determine the weight coefficient corresponding to each dimension of the feature.

[0078] Step S132, the initial risk of each dimension of the feature is calculated according to the set threshold value corresponding to each dimension of the feature, to construct an initial risk vector, and a feature correlation graph is constructed based on the weight coefficient and the initial risk vector.

[0079] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application specifically further includes the following steps: Step S133, the initial risk of each dimension of the feature is iteratively updated by calling a linear diffusion model, and the fused node risk is output after convergence.

[0080] Step S134: Calculate the overall risk index based on the risk of the fusion node corresponding to each dimension feature, and calculate the grid connection readiness score based on the overall risk index.

[0081] The overall risk index is the weighted average of the risks of the fusion nodes corresponding to each dimension of features.

[0082] Step S140: Map the grid connection readiness score to a preset interval to determine the ramp start time window.

[0083] The upper limit of the preset interval is the longest ramp start delay time, and the lower limit is the shortest ramp start delay time.

[0084] In a specific embodiment, the multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system provided by the present invention includes step 3, multi-mode cross-fusion evaluation.

[0085] Includes the following steps: Step 3.1: Construct a feature association graph.

[0086] Specifically, firstly, based on the grid connection feature vector obtained in step 2... Constructing a five-dimensional grid-connected feature vector : ; Again Each component in the graph is used as a feature node to construct a corresponding node set, and then the influence weight of each node is determined based on empirical values. , representing the strength of the influence of node i on node j. In this example, the influence weight can be determined by simulating sensitivity. In the microgrid digital simulation platform, a small perturbation (±5%~10%) is applied to each characteristic parameter, and the rate of change in response to other characteristics or overall grid-connected impact indicators (such as peak current) is observed. Parameters with higher sensitivity are assigned greater weight. Alternatively, hardware-in-the-loop (HIL) testing can be used: on a flywheel-battery system test bench, a phase or frequency step is actually applied, and the changes in power and thermal management for each path are recorded. The quantification effect is compared to determine... .

[0087] After determining the influence weights between each node, the risk of each feature is calculated using the threshold method. The expression is: ; In the formula, Let be the security threshold corresponding to the i-th feature (i.e., the i-th element / node). For the limit value, express The first in There are elements, among which... Therefore, the initial risk .

[0088] Step 3.2, risk propagation fusion.

[0089] Specifically, according to the correlation graph and the initial risk obtained in step 3.1, the risk is iteratively updated using a linear diffusion model, and the expression is: ; In the formula, is the node risk of the t-th iteration, is the attenuation factor, which is 0.85; is the degree matrix, and the diagonal element value of the degree matrix is the sum of the element values corresponding to the row of the weight matrix; is the initial risk, and the iteration converges when , and the fused node risk is output after convergence.

[0090] Step 3.3, comprehensive risk and grid connection readiness score.

[0091] Specifically, the weighted average value of the five-way risk in the fused node risk is calculated, which is the overall risk index , and then the grid connection preparation score is calculated according to the overall risk index : .

[0092] Among them, the average of the fused node risk can reflect the overall system risk in the closing transient state, and the grid connection preparation is its complement, the higher the value, the safer the system.

[0093] Step 3.4, slope start time window mapping.

[0094] Specifically, the grid connection readiness score obtained in step 3.3 is mapped to the interval [50ms, 500ms] to output the slope start time window , and the expression is: ; In the formula, , .

[0095] It should be noted that respectively represent the shortest ramp-up delay and the longest ramp-up delay. Since the short-circuit current impulse during the microgrid closing often reaches the peak value within milliseconds to tens of milliseconds. If the delay is too long, the bus and the protection device may experience multiple overshoots in an uncontrolled manner. Usually, 50 ms is taken as the shortest ramp-up delay, which can not only ensure that the control system completes signal acquisition, calculation and instruction issuance, but also quickly open the ramp schedule before the first short-circuit impact has not fully arrived, avoiding the impact amplitude exceeding the limit. In addition, 500 ms is within the upper limit range of the buffer window commonly used in UPS and flywheel-battery hybrid energy storage systems, which can smooth the transition and will not cause the main grid or load to affect the standby power generation or re-grid operation due to too long isolation.

[0096] In the embodiment, according to the current safety score, the flywheel and battery power stage injection / absorption can be provided with accurate ramp-up time length, which can ensure that the short-circuit impulse is borne step by step and instantaneous over-limit is avoided.

[0097] In step S150, the ramp parameterization processing is performed based on the ramp start time window to construct a power step sequence, and a plurality of control packets issuing instructions according to the power steps are generated according to the power step sequence.

[0098] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the present application specifically includes the following steps: In step S151, the bus rated power capacity is obtained, the rated power peak upper limit is calculated based on the grid-connected preparation degree score and the bus rated power capacity, and the rated power peak upper limit is the product of the grid-connected preparation degree score and the bus rated power capacity.

[0099] In step S152, the ratio of the primary grid injection power to the rated power peak upper limit is set as the primary ramp amplitude ratio, the total number of steps is calculated in combination with the sampling period, and the ramp parameterization processing is performed.

[0100] In some embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the present application specifically includes the following steps: In step S153, the primary power is calculated based on the primary ramp amplitude ratio and the rated power peak upper limit, and the primary time length and the primary step number are calculated according to the ramp start time window.

[0101] In step S154, the power increment of each step in the remaining ramp steps except the primary step number is calculated, and the steps are generated based on the power increment of each step in the remaining ramp steps to construct a power step sequence according to the generated steps.

[0102] In the power step sequence, each power step corresponds to a control packet issuing an instruction of one step.

[0103] In specific embodiments, the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method provided by the application comprises the following steps:

[0104] comprising the following steps: Step 4.1, slope parameterization.

[0105] Specifically, first, according to the bus rated power capacity and the safety margin, the set power peak upper limit is determined: ; In the formula, when , , high safety allows high amplitude grid connection.

[0106] It should be noted that is the maximum output (or energy absorption) power of the flywheel + battery hybrid energy storage system allowed at the grid connection moment, which is used to limit the highest power of the slope sequence to ensure that it does not exceed the bus rated capacity and the limit that the system can withstand.

[0107] Secondly, the proportion of the primary small amplitude grid injection power to the peak upper limit is determined as , wherein the lower limit 0.05 (5% of the peak value) ensures that even a low score can have a minimum response, and the upper limit 0.2 (20% of the peak value) avoids excessive impact in the initial stage.

[0108] Finally, according to the fastest response ability of the control system, the value period is 10ms, so the total number of slopes is the ratio between the slope starting time window and 10ms.

[0109] Step 4.2, primary small amplitude grid power calculation.

[0110] Specifically, first, the primary power , the primary time length and the corresponding primary step length are calculated: ; ; ; In the formula, if =0.1, =500kW, then =50kW, when =200ms, =20ms.

[0111] This step injects / absorbs power at a small amplitude in a very short time, withstands the initial pulse, and gains a buffer time for the subsequent larger power cut-in.

[0112] Step 4.3, power step generation.

[0113] Specifically, first, for the remaining ramp except the primary step, the number of steps is M=N- The power of each step is increased by: ; Second, generate steps based on the power of each step of the remaining ramp: ; In the formula, represents the power of the i-th step, N represents the total number of steps of the ramp, and finally needs to ensure .

[0114] This step gradually increases the power from a small amplitude to a peak value through equal-amplitude power step increments, which can smoothly transition and fully utilize the score and grid readiness.

[0115] Step 4.4, instruction issuance and closed-loop monitoring.

[0116] Specifically, first, generate the control package of each step in the step according to the power increment step obtained in step 4.3, denoted as: ; In the formula, represents the time stamp, represents the time stamp of grid connection, is the interval time of the timing instruction.

[0117] Then, send the first small amplitude injection instruction {0, } to the flywheel and battery converter through the EtherCAT bus, and then send the power instruction in turn according to the timing. Synchronously collect the bus voltage and actual output power, and if a sudden change or protection action warning occurs, trigger an emergency stop or back up to the previous step of the ladder.

[0118] This step gradually executes the designed power step scheme, which can ensure the execution effect through real-time monitoring, and forms a closed-loop scheduling together with the aforementioned grid readiness score and ramp window.

[0119] The flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling system provided by the present application is described below. The flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling system described below can be mutually corresponding and referred to the flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling method described above.

[0120] As Figure 2As shown, in one embodiment, a flywheel and battery hybrid energy storage system multi-mode cross data fusion scheduling system includes a data acquisition and fusion module, a feature extraction module, a grid-connected readiness score module, a ramp window determination module, and an instruction issuing module.

[0121] The data acquisition and fusion module is configured to acquire multi-mode signals in parallel, and fuse and package the multi-mode signals into a unified time sequence multi-mode data vector.

[0122] The feature extraction module is configured to pre-process and normalize the multi-mode data vector to obtain a pre-processed sequence, and extract multi-dimensional grid-connected features from the pre-processed sequence.

[0123] The grid-connected readiness score module is configured to construct a feature correlation graph based on the multi-dimensional grid-connected features, and fuse the node risks of the feature correlation graph through a linear diffusion model to determine a grid-connected readiness score.

[0124] The ramp window determination module is configured to map the grid-connected readiness score to a preset interval to determine a ramp start time window.

[0125] The instruction issuing module is configured to perform ramp parameterization processing based on the ramp start time window to construct a power step sequence, and generate a plurality of control packets according to the power step sequence, wherein the control packets are configured to issue instructions according to the power steps.

[0126] In the above embodiment, the multi-mode signals include bus frequency and phase signals on a public AC bus, flywheel output power on an AC side of each flywheel energy storage converter, battery state of charge and maximum discharge power, and cooler output power; and the upper limit of the preset interval is the longest ramp start delay time, and the lower limit of the preset interval is the shortest ramp start delay time.

[0127] The present disclosure can be a system, a method, and / or a computer program product. The computer program product can include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present disclosure.

[0128] Computer readable storage media can be tangible storage media which can retain and store instructions for use by an instruction execution device. Computer readable storage media can be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer readable storage media include the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.

[0129] Computer readable program instructions described herein can be downloaded to respective computing / processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network can comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing / processing device.

[0130] Computer readable program instructions for carrying out operations of the present disclosure can be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate array (FPGA), or programmable logic array (PLA) can execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.

[0131] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0132] These computer readable program instructions can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer readable program instructions can also be stored in a computer readable storage medium that can include random access memory (RAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other data storage device. When the computer readable program instructions are loaded into the computer and other programmable data processing apparatus, a series of operational steps are implemented that provide processes such that the computer or other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0133] The computer readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable data processing apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

Claims

1. A multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system, characterized in that, The method includes: Parallel acquisition of multi-mode signals, and fusion and packaging of the multi-mode signals into a unified time-series multi-mode data vector; the multi-mode signals are the bus frequency and phase signals on the common AC bus, the flywheel output power on the AC side of each flywheel energy storage converter, the battery state of charge and maximum discharge power, and the cooler output power; The multi-mode data vector is preprocessed and normalized to obtain a preprocessed sequence, and multi-dimensional grid-connected features are extracted from the preprocessed sequence. Based on the multidimensional grid connection characteristics, a feature association graph is constructed, and the node risks of the feature association graph are fused through a linear diffusion model to determine the grid connection readiness score. The grid connection readiness score is mapped to a preset interval to determine the ramp start time window; the upper limit of the preset interval is the longest ramp start delay time, and the lower limit is the shortest ramp start delay time. Based on the slope start time window, slope parameterization is performed to construct a power step sequence, and multiple control packets are generated according to the power step sequence to issue commands according to the power step.

2. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 1, characterized in that, The parallel acquisition of multi-mode signals, and the fusion and packaging of the multi-mode signals into a unified time-series multi-mode data vector, includes: A frequency transmitter and a synchronous phase measurement unit are installed on the common AC bus, and the bus frequency and phase signals with timestamps are collected by using a synchronous sampling method combined with a GPS clock. Voltage and current sensors are installed on the AC side of each flywheel energy storage converter. The power calculation unit executes the power calculation formula and collects the flywheel output power with timestamps.

3. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 2, characterized in that, The parallel acquisition of multi-mode signals, and the fusion and packaging of the multi-mode signals into a unified time-series multi-mode data vector, further includes: The SOC is updated in real time in the battery management unit by coulomb measurement and the maximum discharge power of the battery is estimated by looking up a table based on the battery temperature and internal resistance. Combined with the CAN or EtherCAT bus, the state of charge and maximum discharge power of the battery are obtained. Flow meters and temperature difference sensors are installed in the cooling circuits of the flywheel nacelle and battery cabinet to calculate the cooling power in real time and determine the output power of the cooler based on the cooling power. The host computer synchronizes the multi-mode signal with a clock, resamples and interpolates it at a preset frequency, and fuses and packages the multi-mode signal to obtain the multi-mode data vector.

4. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 1, characterized in that, The preprocessing and normalization of the multi-mode data vector to obtain a preprocessed sequence, and the extraction of multi-dimensional grid-connected features from the preprocessed sequence, include: The central control unit performs a mutation threshold test on each component of the multimodal data vector to mark and remove outlier data, and performs linear interpolation to fill in missing values. The multi-modal data vector after the abnormal data and missing values ​​are filled is subjected to noise filtering and smoothing, and then resampled using a unified time grid to normalize the multi-modal data vector, resulting in a normalized preprocessed sequence.

5. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 4, characterized in that, The step of preprocessing and normalizing the multi-mode data vector to obtain a preprocessed sequence, and extracting multi-dimensional grid-connected features from the preprocessed sequence, further includes: The bus frequency sequence is extracted from the preprocessed sequence, and the frequency change rate is calculated in the real-time processor at fixed time intervals to obtain the frequency change rate sequence. The phase signal sequence is extracted from the preprocessed sequence, and the absolute value of the phase difference at each sampling point in the phase signal sequence is calculated to construct an absolute phase deviation sequence.

6. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 5, characterized in that, The step of preprocessing and normalizing the multi-mode data vector to obtain a preprocessed sequence, and extracting multi-dimensional grid-connected features from the preprocessed sequence, further includes: The flywheel output power sequence is extracted from the preprocessed sequence, and the flywheel output power increment is calculated by the controller at the same time intervals. The battery state of charge and maximum discharge power sequence are extracted from the preprocessed sequence, and the battery state of charge and maximum discharge power are fused in the BMS to obtain the remaining usable power index of the battery. Extract the cooler output power sequence from the preprocessed sequence and calculate the cooling margin coefficient based on the rated cooling power of the cooling unit.

7. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 1, characterized in that, The process of constructing a feature association graph based on the multi-dimensional grid connection characteristics, and fusing the node risks of the feature association graph using a linear diffusion model to determine the grid connection readiness score, includes: Based on the multi-dimensional grid connection characteristics, a multi-dimensional feature vector is constructed, and the features of each dimension are perturbed in the microgrid digital simulation platform to determine the weight coefficients corresponding to each dimension. The initial risk of each feature is calculated according to the set threshold corresponding to each feature, so as to construct an initial risk vector, and the feature association graph is constructed based on the weight coefficient and the initial risk vector.

8. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 7, characterized in that, The step of constructing a feature association graph based on the multi-dimensional grid connection characteristics, and fusing the node risks of the feature association graph using a linear diffusion model to determine the grid connection readiness score, further includes: The linear diffusion model is invoked to iteratively update the initial risk of each feature dimension, and the risk of the fusion node is output after convergence. The overall risk index is calculated based on the risk of the fusion node corresponding to each dimension of the characteristics, and the grid connection readiness score is calculated based on the overall risk index. The overall risk index is the weighted average of the risks of the fusion nodes corresponding to each of the features.

9. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 1, characterized in that, The process of parameterizing the ramp based on the ramp start time window to construct a power step sequence, and generating multiple control packets according to the power step sequence to issue commands according to the power step, includes: Obtain the rated power capacity of the bus, and calculate the upper limit of the rated power peak based on the grid connection readiness score and the rated power capacity of the bus, wherein the upper limit of the rated power peak is the product of the grid connection readiness score and the rated power capacity of the bus; The ratio of the primary grid-injected power to the upper limit of the rated power peak is set as the primary ramp amplitude ratio, and the total number of ramp steps is calculated in combination with the sampling period to perform ramp parameterization processing.

10. The multi-mode cross-data fusion scheduling method for a flywheel and battery hybrid energy storage system according to claim 9, characterized in that, The step of performing slope parameterization processing based on the slope start time window to construct a power step sequence, and generating multiple control packets according to the power step sequence to issue commands according to the power step, further includes: The primary power is calculated based on the primary ramp amplitude ratio and the upper limit of the rated power peak, and the primary duration and primary steps are calculated based on the ramp start time window. The power increment of each step in the remaining ramp steps other than the initial steps is taken, and a step is generated based on the power increment of each step in the remaining ramp steps, so as to construct the power step sequence according to the generated step. In the power ladder sequence, each power ladder corresponds to a control packet that issues a command at one step.

Images