A system and method for suppressing resonance in a parallel high-capacity capacitor bank

Abstract

The application discloses a system and method for suppressing parallel large-capacity capacitor bank resonance. The system obtains a series resonance candidate frequency point set and a parallel resonance candidate frequency point set through offline identification before operation or in a maintenance debugging stage, and determines damping module parameters; in an online running stage, a control unit performs online preprocessing and composite resonance risk index calculation around the candidate frequency points, controls corresponding damping modules to be put into use in a first level, a second level or in combination, maintains input and cuts off according to the series resonance risk index and the parallel resonance risk index. The application can selectively put damping modules into use according to different resonance types, candidate frequency points and sensitive partitions, reduce current amplification, node voltage amplification and partition circulating current risk of the parallel capacitor bank in a target frequency band, and reduce additional loss caused by long-term input of the damping branch.

Description

Technical Field

[0001] This invention pertains to electrical connection and oscillation suppression technology in power electronics and high-power energy storage / DC bus systems, and particularly relates to a system and method for suppressing resonance of parallel large-capacity capacitor banks. Background Technology

[0002] In high-voltage DC bus, high-power energy storage, pulsed power devices, and high-ripple current filtering scenarios, it is often necessary to connect multiple large-capacity capacitors (such as DC support capacitors / filter capacitors / energy storage capacitors) in parallel between the positive and negative busbars to obtain the required total capacitance and ripple current carrying capacity. Due to the inevitable introduction of parasitic inductance, parasitic resistance, and mutual coupling by the busbar length, connecting pieces, terminals, capacitor lead-out structures, and spatial arrangement, multiple capacitor branches will form a multi-node, multi-modal distributed parameter network.

[0003] Within a specific frequency range, this network may exhibit two types of risks: First, local valleys in the equivalent impedance can amplify external excitation or internal disturbance currents at the corresponding frequency points, forming strong circulating currents between partitions, exhibiting a series resonance risk. Second, local peaks in the equivalent impedance can cause node voltages to respond significantly to small harmonic or disturbance currents, exhibiting a parallel resonance risk. These risks can lead to increased capacitor ripple current, overheating at connections, node voltage overshoot, reduced device lifetime, and even system instability.

[0004] In existing technologies, common vibration suppression methods include fixed damping resistors, RC damping branches, RLC resonant absorbers, active damping, or controller compensation. However, for parallel capacitor banks, resonance often has the characteristics of "multi-frequency points, partitioning, and location dependence." If the damping branch is fixedly connected for a long period of time, it may cause additional losses and thermal burden; if only global current or voltage thresholds are relied upon, it is difficult to distinguish between series resonance and parallel resonance, and it is also difficult to decide which type of damping module should be connected, which partition should be connected, and when to disconnect it. Summary of the Invention

[0005] The technical problem to be solved by this invention is how to suppress the resonance of parallel large-capacity capacitor banks in a specific frequency band (500–2000 Hz) while reducing the additional losses of the damping branch.

[0006] This invention solves the above-mentioned technical problems through the following technical solution: A system for suppressing resonance of a parallel large-capacity capacitor bank, comprising: a parallel capacitor bank, a busbar system, a partitioned two-stage busbar structure, a damping module, a sampling unit, a control unit, and a damping switching drive unit; the parallel capacitor bank includes at least two capacitor partitions, each capacitor partition being connected to a positive busbar and a negative busbar via a partitioned busbar; the damping module includes multiple damping modules disposed at different partition nodes and / or key busbar nodes, each damping module being connected to a corresponding node via an independent switching device; the sampling unit is used to collect bus voltage, total current, and At least a portion of the partition current and / or partition node voltage; the control unit pre-stores the series resonance candidate frequency set and the parallel resonance candidate frequency set obtained by offline identification, and performs online preprocessing and resonance feature extraction on the candidate frequencies to form series resonance risk indicators and parallel resonance risk indicators; when the corresponding risk indicator exceeds the input threshold and continues to meet the preset time, the damping switching drive unit controls at least one damping module corresponding to the resonance type, candidate frequency point and sensitive partition to be connected; when the corresponding risk indicator is lower than the release threshold and continues to meet the preset time, the at least one damping module is controlled to be disconnected.

[0007] As a preferred technical solution, the damping module includes at least one of the following: an RC series damping branch, a series RLC absorber branch, and a low-pass damping branch; the low-pass damping branch includes a parallel unit formed by a damping resistor and a parallel reactor, and the parallel unit is connected in series with a damping capacitor and then connected to the corresponding partition node.

[0008] As a preferred technical solution, the damping modules are not all simultaneously engaged under normal conditions, but are selectively operated in at least one of the following ways: single damping module engagement, multi-damping module engagement at the same frequency, coordinated engagement of damping modules in adjacent zones, and combined engagement under multi-frequency events; only when multiple candidate frequency points are triggered simultaneously or the resonance risk index reaches a severe threshold are all or most of the damping modules allowed to be engaged.

[0009] As a preferred technical solution, the control unit performs preprocessing of candidate frequency point constraints on each candidate frequency point. The preprocessing includes at least: removing DC components, establishing an adaptive narrowband window around the candidate frequency point, calculating bandpass energy within the sliding time window, and using the Goertzel algorithm, single-frequency discrete Fourier analysis, or equivalent single-frequency amplitude estimation algorithm to obtain the voltage amplitude, current amplitude, and energy characteristics of the corresponding candidate frequency point.

[0010] As a preferred technical solution, the series resonance risk index is composed of at least the current amplification factor, impedance valley factor, partitioned circulating current factor, and energy growth rate factor after weighted normalization; the parallel resonance risk index is composed of at least the node voltage amplification factor, impedance peak factor, disturbance sensitivity factor, and energy growth rate factor after weighted normalization.

[0011] As a preferred technical solution, when the control unit determines that the series resonance characteristics are strong, it prioritizes activating the RC damping branch and / or the series RLC absorber branch located at the node with the largest partitioned circulating current factor or the largest current amplification factor; when it determines that the parallel resonance characteristics are strong, it prioritizes activating the series RLC absorber branch and / or the low-pass damping branch located at the node with the largest node voltage amplification factor or the largest impedance peak factor. This is done when the series resonance risk index is ≥ 1 and at least two of the series resonance criteria are simultaneously met and remain in effect for a certain period. T on When the series resonance risk index is 0.70 ≤ 1, it is judged as having strong series resonance characteristics. When the series resonance risk index is ≥ 1 and at least two of the parallel resonance criteria are met simultaneously and continuously, it is judged as having weak series resonance characteristics or being in a warning state. T on When the value is 0.70, it is judged to have strong parallel resonance characteristics; when the value is 0.70 ≤ parallel resonance risk index < 1, it is judged to have weak parallel resonance characteristics or be in a warning state.

[0012] As a preferred technical solution, the series resonant candidate frequency point set is obtained by the occurrence of local impedance valleys and the accompanying current amplitude amplification during the offline frequency sweep process; the parallel resonant candidate frequency point set is obtained by the occurrence of local impedance peaks and the accompanying increase in node voltage response during the offline frequency sweep process.

[0013] As a preferred technical solution, the offline identification also includes a ringing test. Before the system is put into operation or during the maintenance and debugging phase, a step excitation or pulse excitation is applied to the parallel capacitor bank and busbar network, and their damped oscillation response is observed. The control unit estimates the ringing frequency based on the interval between adjacent peaks in the ringing waveform, and estimates the damping ratio and equivalent natural frequency based on the attenuation relationship of adjacent peak amplitudes. The ringing test results are used to verify, supplement or correct the parallel resonance candidate frequency point set and series resonance candidate frequency point set obtained by offline frequency sweep.

[0014] As a preferred technical solution, the parameters of the damping module are tuned based on the confirmed target frequency identified offline: the resonant frequency of the series RLC absorber branch is close to the confirmed target frequency; the cutoff frequency of the RC series damping branch is near the confirmed target frequency; and the parameters of the parallel reactor in the low-pass damping branch satisfy both low-frequency bypass and the confirmed target frequency. The dual constraints of energy consumption.

[0015] As a preferred technical solution, the damping module parameter tuning is also based on a comprehensive objective function that takes into account both resonance suppression effect and additional loss. The comprehensive objective function includes at least two or more of the following: impedance peak reduction ratio, impedance valley rise ratio, damping branch power consumption ratio, and damping branch current margin.

[0016] As a preferred technical solution, the damping module is arranged at at least one of the following types of nodes: the connection node between the partition busbar and the positive main busbar, the connection node between the partition busbar and the negative main busbar, the coupling node between adjacent partitions, the busbar transition node, the impedance change node, the middle node of the main busbar, and the end node of the main busbar.

[0017] The present invention also provides a method for suppressing resonance in a parallel large-capacity capacitor bank system as described in any of the above-mentioned schemes, comprising the following steps: S1, offline frequency sweep identification and / or ringing test are used to obtain the candidate frequency points of series resonance and the candidate frequency points of parallel resonance; S2, online acquisition of bus voltage, total current, zone current and / or zone node voltage; S3 performs DC removal, adaptive narrowband filtering, sliding window energy calculation, and single-frequency amplitude estimation around the candidate frequency point; S4, construct series resonance risk indicators and parallel resonance risk indicators; S5. Selectively deploy at least one damping module based on the magnitude, duration and location of the risk indicator and the sensitive zone. S6, after the risk indicator is lower than the release threshold and continues to meet the preset time, the corresponding damping module is cut off; S7, based on the changes in impedance peak and valley, branch current and node voltage before and after the damping is put into operation, the damping parameters are re-measured and adjusted.

[0018] This invention provides an integrated solution combining structure, identification, control, and tuning. Structurally, it includes a parallel capacitor bank with partitioned secondary busbars, multiple distributed damping modules, a sampling unit, a control unit, and a damping switching drive unit; methodologically, it includes offline candidate frequency identification, online candidate frequency constraint preprocessing, composite resonance risk index calculation, selective activation / deactivation of damping modules, and damping parameter retesting and tuning.

[0019] Compared to simple fixed damping, this invention can only activate the corresponding damping branch when the target resonance event occurs, reducing long-term additional losses. Compared to schemes that rely solely on a single current or voltage threshold, this invention uses a composite risk index, which can reduce false positives and false negatives, and distinguish between series resonance risk and parallel resonance risk. The candidate frequency point set serves both online detection and directly the design and retesting of damping branch parameters, forming a closed loop of "identification-design-verification". The selective activation of multiple damping modules according to their location and type is beneficial for suppressing multi-zone and multi-modal resonance, while avoiding unnecessary global energy consumption. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall structure of the parallel capacitor bank, the two-stage busbar, the three switchable damping modules, and the control unit according to an embodiment of the present invention.

[0021] Figure 2 This is a flowchart illustrating the entire operation process of a system for suppressing resonance of a parallel large-capacity capacitor bank, as described in an embodiment of the present invention. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. Equivalent substitutions or modifications made by those skilled in the art to the structural form, device parameters, number of partitions, damping module form, and control logic without departing from the concept of the present invention should all fall within the scope of protection of the present invention.

[0023] The "parallel large-capacity capacitor bank" described in this invention is applied to systems such as DC bus support, pulse power storage, filtering, and energy buffering. The parallel large-capacity capacitor bank comprises multiple individual capacitors, each of which is a film capacitor, power electronic capacitor, energy storage capacitor, or other capacitor components suitable for high-voltage, high-current scenarios.

[0024] The term "large capacity" in this application primarily refers to the large total equivalent capacity of a capacitor bank formed by connecting multiple individual capacitors in parallel, rather than being judged solely by the capacity of a single capacitor. The rated capacity of a single capacitor is 1mF to 100mF, preferably 10mF to 20mF; the total equivalent capacity of a parallel capacitor bank is 10mF to 10F, preferably 100mF to 0.5F. The total equivalent capacity is determined by the sum of the rated capacities of all capacitors connected in parallel. The specific number of capacitors, their individual capacities, and the total equivalent capacity are determined based on the system's rated voltage, energy storage requirements, filtering requirements, and ripple current carrying capacity requirements.

[0025] The busbar system can be a laminated busbar or a non-laminated busbar; the damping module can be composed of passive devices, or combined with switching devices to form a switchable damping unit. Figure 1 The three damping modules are equipped with independent switching devices. During normal operation, not all of them are required to be engaged. Instead, the control unit selectively engages them based on the resonance type, candidate frequency point, and sensitive zone.

[0026] In this embodiment, multiple capacitor cells are divided into M capacitor zones, where M is not less than 2. Each capacitor zone includes at least two capacitor cells connected in parallel. Each capacitor zone first forms a local parallel connection within its own zone, and then connects to the positive main busbar and the negative main busbar via the corresponding zone busbar, forming a zoned two-stage busbar structure. The equivalent capacity of each capacitor zone is determined based on the number of capacitor cells it contains and the rated capacity of each capacitor cell. Different capacitor zones may have the same equivalent capacity, or they may be set to different capacities according to system layout and current distribution requirements. Figure 1 As shown, the parallel large-capacity capacitor bank includes capacitor bank A, capacitor bank B, and capacitor bank C. Each of capacitor bank A, capacitor bank B, and capacitor bank C is formed by multiple individual capacitors connected in parallel, and is connected to the positive main busbar and the negative main busbar respectively through corresponding partition busbars.

[0027] Multiple damping modules are installed at each partition node, busbar transition node, busbar middle node, end node, or impedance change node. For example... Figure 1 As shown, the partition nodes refer to the positions where capacitor banks A, B, and C are respectively connected to the positive and negative main busbars; the busbar transition nodes refer to the positions where the partition busbars and main busbars are electrically connected via connecting pieces, transition pieces, bolt crimping parts, copper busbar bending sections, quick-connect parts, etc. Figure 1 The positions of the two ends of the capacitor bank in the middle section; the middle node of the busbar refers to the middle area along the length of the positive main busbar and the negative main busbar. If the number of individual capacitors in capacitor banks A, B, and C is the same, the middle node of the busbar is located at... Figure 1 The location of the node is the same as the partition node of capacitor bank B; the end node refers to the location near both ends of the positive and negative main busbars, including the location near the outermost capacitor partition and the busbar ends far from the main input terminal, where path impedance differences are easily formed; the impedance abrupt change node refers to the location where the width, thickness, material, connection method, and current carrying path of the busbar change significantly. The specific location of the impedance abrupt change node is determined based on the busbar structure, connection method, device arrangement, and offline identification results.

[0028] The damping modules are connected to corresponding nodes via independent switching devices, allowing different damping modules to be in either off, primary activation, secondary additional activation, or sustained activation states. Primary activation refers to activating the most suitable first damping module after an initial assessment of resonance risk at a candidate frequency. Secondary additional activation refers to continuing to observe risk indicators after primary activation; if the resonance risk decreases insufficiently after the evaluation period (e.g., less than 20%), a second damping module is activated while the primary activation remains in place. Sustained activation refers to maintaining the activated damping module's connection even after the risk indicator has decreased but the release condition has not yet been met; no new damping modules are added until the risk indicator falls below the release threshold and remains below it for a set time before the corresponding damping module is deactivated.

[0029] In this embodiment, the damping module may take one or more of the following forms.

[0030] 1) RC series damping branch: The damping resistor Rd and the damping capacitor Cd are connected in series to the node to be damped. At low frequencies, the capacitive reactance of Cd is relatively high, and the branch current is limited; near the target frequency band, the capacitive reactance of Cd decreases, and the resonant energy can be dissipated through Rd.

[0031] 2) Series RLC absorber branch: The damping resistor Rd, damping inductor Ld, and damping capacitor Cd are connected in series to the node to be suppressed. By tuning this branch to near the confirmed target frequency, it forms an energy absorption channel in that frequency band.

[0032] 3) Low-pass damping branch: The damping resistor Rd and the shunt reactor Lb are connected in parallel to form a parallel unit. This parallel unit is then connected in series with the damping capacitor Cd and connected to the target node. At low frequencies, Lb provides a bypass, reducing the continuous loss of Rd; confirm the target frequency. As the internal impedance of Lb increases, more resonant current flows through Rd and dissipates.

[0033] The present invention provides a method for suppressing resonance of parallel large-capacity capacitor banks. This method is applied to a system for suppressing resonance of parallel large-capacity capacitor banks. The system includes parallel capacitor banks, busbar system, partitioned secondary bus structure, damping module, sampling unit, control unit, and damping switching drive unit. The sampling unit (not shown) is used to collect one or more of the following signals: bus voltage, total current, zone current, zone node voltage, damping branch current, damping module temperature, and switching status. The control unit uses the sampled signals to identify resonance at candidate frequency points and make damping switching decisions. The damping switching drive unit controls the switching devices according to the instructions of the control unit, switching the damping module between open, primary-level engaged, secondary-level additional engaged, or maintained engaged states.

[0034] See Figure 2As shown, this invention illustrates the entire process of suppressing the resonance of a parallel large-capacity capacitor bank system before commissioning or during maintenance and commissioning, including offline frequency sweep identification, optional ringing test, damping parameter design and retest verification, and online preprocessing of candidate frequency point constraints, composite resonance risk assessment, first-level input of damping modules, second-level or combined input, release and removal, and cyclic monitoring.

[0035] The method of the present invention for suppressing resonance of parallel large-capacity capacitor banks includes the following steps: Step 1) Identify candidate frequency points for series resonance and parallel resonance within the target frequency band of 500–2000 Hz, specifically including: To ensure that online detection and damping tuning have a clear target, offline frequency sweep identification is preferred before system commissioning or during maintenance and debugging. A low-amplitude sinusoidal excitation is used to sweep the frequency point-by-point or continuously within the 500–2000 Hz range. The excitation signal can be expressed as:

[0036] in: Inject an excitation signal to the frequency sweep. U inj For the sweep frequency excitation amplitude, t For time, This is the current frequency sweep point; At each frequency point At the selected node, the node voltage amplitude is measured. U ( ), and simultaneously measure the total current amplitude. I ( and the current amplitude of each zone I j ( The equivalent impedance is calculated. The node to be evaluated is a candidate node used for arranging damping modules or for determining the resonant response, selected from at least one of partition nodes, busbar transition nodes, busbar middle nodes, end nodes, and impedance abrupt change nodes. When sweeping frequencies for multiple nodes to be evaluated, the control unit repeats the above measurements and calculations for each node to be evaluated; when determining the candidate frequency point of parallel resonance, U ( Prioritize selecting the voltage amplitude of the node to be evaluated with the largest voltage response at the corresponding frequency point.

[0037]

[0038] Among them, Z ( (Frequency point) The equivalent impedance at that point, e ITo prevent the use of excessively small positive numbers in the denominator and avoid impedance calculation errors caused by excessively small current amplitudes, the following parameters are determined based on the current sampling range and noise level:

[0039] in I ref The reference current amplitude in the frequency sweep test or the reference amplitude corresponding to the system's rated current is σI, where σI is the root mean square value of the current sampling noise. Take values ​​from 0.001 to 0.01; k I Take 3–5.

[0040] Current amplitude of each zone Ij( It does not directly participate in the equivalent impedance Z ( The calculation is not for the partition current dispersion, but for the partition circulation current dispersion and the location of the resonant sensitive partition, which serves as the basis for subsequent selection of the damping module access node:

[0041] in, M This represents the total number of capacitor partitions. This represents the average current across all zones. I cir ( ) is a characteristic quantity of regional circulation. j sen ( This is the sensitive zone with the greatest current deviation.

[0042] If a certain frequency point Satisfy Z ( () is a local maximum value in the neighborhood, and is relative to the neighborhood background impedance. The ratio reaches the first preset threshold. Then, this frequency point is recorded as the candidate frequency point for parallel resonance:

[0043] in, For the first q One frequency point to be determined Z med ( The background impedance in the neighborhood of this frequency point is determined by the median value of the neighborhood impedance amplitude. or p The first preset threshold for judging candidate frequencies of parallel resonance.

[0044] Preferably, the first preset threshold A value of 1.20–1.50 is acceptable, with 1.25 being more preferred.

[0045] If a certain frequency point satisfy It is the local minimum in the neighborhood, and the ratio of the neighborhood background impedance to the impedance at this frequency point reaches the second preset threshold. or s And the current amplitude at that frequency point relative background current If there is amplification, then this frequency point is recorded as a candidate frequency point for series resonance:

[0046]

[0047] in I med ( The background current in the neighborhood of this frequency point is determined by the median value of the neighborhood current amplitude. Z med ( The background impedance in the neighborhood of this frequency point is determined by the median value of the neighborhood impedance amplitude. or s The second preset threshold for determining the valley value of series resonant impedance. or I The third preset threshold is used for current amplification judgment.

[0048] Preferably, the second preset threshold The value can be 1.20–1.50, and more preferably 1.25; the third preset threshold A value of 1.20–1.60 is acceptable, with 1.30 being more preferred.

[0049] Series resonant candidate frequency set obtained by frequency sweep F s Candidate frequency set of parallel resonance F p It is not only used to "mark" resonance risks, but also has at least the following uses: 1. Adaptive narrowband for limiting candidate frequency neighborhood during online detection B q And Goertzel / single-frequency analysis frequency points; After offline frequency scanning, if resonances are found to be prone to occur around 800 Hz and 1250 Hz, the online operation will no longer average the detection across the entire 500–2000 Hz band, but will instead focus on detecting a small frequency band around 800 Hz and 1250 Hz. This reduces computational load and improves detection sensitivity. Voltage, current, and energy changes can be directly calculated for these candidate frequency points using Goertzel or single-frequency analysis algorithms.

[0050] 2. Used to determine the tuning frequency of the series RLC absorber branch; If offline frequency scanning reveals a frequency point prone to resonance, such as 1000 Hz, then the series RLC absorber should be designed with parameters based on the 1000 Hz range. In this way, the series RLC absorber can form a strong absorption channel near that frequency, diverting the resonant energy at that frequency into the damping resistor for dissipation.

[0051] 3. Used to determine the cutoff frequency range of the RC series damping branch; The purpose of an RC series damping branch is not to dissipate a large amount of energy at all frequencies, but rather to achieve high energy dissipation at a specific target frequency. The damping effect becomes apparent in the vicinity. If the candidate resonant frequency is around 1000 Hz, the operating frequency band of the RC series damping branch should be set around 1000 Hz to minimize losses during normal operation at low frequencies, after confirming the target frequency. The interior can provide damping.

[0052] 4. Used to determine the target energy dissipation frequency band for low-pass damping branches; The low-pass damping branch needs to distinguish the normal low-frequency components and confirm the target frequency. Candidate frequencies are used to determine which frequency band a device primarily consumes power in. For example, normal low-frequency components need to be allowed to pass through or bypassed as much as possible to avoid long-term losses; while candidate resonant frequencies identified within the 500–2000 Hz range require the damping resistor to participate in power consumption.

[0053] 5. Used for frequency rescan verification after damping tuning.

[0054] After the damping module parameters are determined and connected, the system undergoes another frequency sweep. The purpose of this re-sweep is to check whether the original candidate resonant frequencies have been effectively suppressed. If the effect is not significant, the damping module parameters, connection position, or switching strategy are adjusted, and then the verification is performed again.

[0055] Steps 2) and 3) are used to transform phenomena such as "significantly amplified current, impedance valley, increased zonal circulating current, significantly amplified node voltage, impedance peak, and increased disturbance sensitivity" into executable and quantifiable criteria. Specifically, step 2) is used to perform online preprocessing of the sampled signal with candidate frequency constraints and extract feature quantities, and step 3) is used to construct a composite resonance risk index and quantitative criteria based on the feature quantities.

[0056] Step 2) Online preprocessing of candidate frequency constraints. During online operation, the control unit does not perform equal weighting on the entire frequency range of 500–2000 Hz, but instead focuses on the set of candidate frequency points for series resonance. F s Candidate frequency set of parallel resonanceF p Candidate frequency points are preprocessed to reduce computational load and enhance targeting accuracy. Specifically, the following steps are included: S1: Remove the DC component. For any original discrete sampling sequence x [ n The moving average differential method is used to remove DC and slow drift.

[0057] in, x [ n [] is the original discrete sampling sequence, representing any sampled signal among bus voltage, total current, zone current, or zone node voltage; x 0[ n This is the sampling sequence after removing the DC component and slow drift; n This is the current sampling point number; m This refers to the sampling point number used in the summation calculation of the moving average; N 0 represents the DC estimation window length.

[0058] S2: Establish an adaptive narrowband neighborhood around the candidate frequency points B q For candidate frequency points (including the candidate frequency set around series resonance) F s Candidate frequency set of parallel resonance F p Define candidate frequency neighborhood adaptive narrowband Bq for:

[0059] in, This is half the bandwidth of the adaptive narrowband; For minimum bandwidth, r The relative bandwidth factor is preferably 3%–10%.

[0060] S3: In relation to candidate frequency points The bandpass energy is calculated within the corresponding sliding time window. This applies to adaptive narrowband applications in the candidate frequency neighborhood. B q Filtered signal Define bandpass energy :

[0061] in, x Bq [ n [The original discrete sampling sequence] x [ nThe sampling sequence is obtained by DC removal and then adaptive narrowband Bq filtering of candidate frequency point neighborhood. For signal x [ n In candidate frequency points Nearby, sliding time window k The bandpass energy within; k This is the sequence number of the last sampling point of the current sliding time window; N w Calculate the window length for bandpass energy; T w,q To and N w The corresponding time window is selected to cover candidate frequencies. 4–12 cycles.

[0062] S4: For cases where the number of candidate frequency points is small, the control unit uses the Goertzel algorithm, single-frequency discrete Fourier analysis, or equivalent single-frequency amplitude estimation algorithm to extract candidate frequency points. voltage amplitude at U q Total current amplitude I q Zone current amplitude I j,q and bandpass energy E q ( k ), used for subsequent risk indicator calculations.

[0063] Step 3) Composite Resonance Risk Indicators and Quantitative Criteria To overcome the problem of misjudgment caused by a single threshold, this step further proposes a composite resonance risk index. This index is a fusion criterion designed in this application for the resonance suppression scenario of parallel large-capacity capacitor banks. It does not only use a single RMS value or a single impedance peak and valley, but also performs a weighted evaluation after unifying and normalizing multiple types of physical characteristics.

[0064] For any candidate frequency point and the current sliding window k Constructing a series resonance risk index Φ s ( , k ) and parallel resonance risk index Φ p ( , k ):

[0065]

[0066]

[0067]

[0068] The meanings of each parameter are shown in Table 1 below: Table 1 Parameter Meaning

[0069] The above factors can be specifically defined as:

[0070]

[0071]

[0072]

[0073]

[0074]

[0075]

[0076]

[0077]

[0078] in, I q Candidate frequency points The total current amplitude at the location; I q,0 Candidate frequency points The reference value of the total current at the location; U q Candidate frequency points The magnitude of the node voltage at that location; U q,0 Candidate frequency points The node voltage reference value at the location; Z q Candidate frequency points The current equivalent impedance at the point; Z q,0 Candidate frequency points The equivalent impedance reference value at the location; I j , q For the first j Each capacitor partition at the candidate frequency point The amplitude of the partition current at the location; j This refers to the capacitor partition number; M This represents the total number of capacitor partitions. Candidate frequency points The average value of the current amplitude in each zone; I cir,q Candidate frequency points The characteristic quantity of the regional circulation at the location; E q ( k ) as candidate frequency points In the current sliding time window k The bandpass energy within; E q ( k -1) is a candidate frequency point The bandpass energy within the previous sliding time window; Δ U q Candidate frequency points The change in the node voltage magnitude relative to the reference state or the previous time window; Δ I q Candidate frequency points The change in the total current amplitude relative to the reference state or the previous time window; the subscript 0 indicates the health reference value, the factory commissioning reference value, or the rolling reference value before the current event is triggered. e I , e U , e Z , e E These are dimensionless zero-prevention terms for current, voltage, impedance, and energy, respectively. e C and e S For dimensionless zero-terms.

[0079] To address the quantitative requirements of expressions such as "significant amplification," "increased circulating current," "significant amplification of node voltage," and "increased sensitivity to disturbances," this invention preferably adopts the judgment criteria in Table 2 below, but is not limited thereto: Table 2 Judgment Criteria

[0080] In a preferred embodiment, when Φ s ( , k ) ≥ 1 and at least two of the above-mentioned series resonance-related criteria are simultaneously satisfied and persist. T on When 0.70 ≤ Φ, it is judged to have strong series resonance characteristics; s ( , k When ) < 1, it can be judged that the series resonance characteristics are weak or that it is in a warning state.

[0081] T onCandidate frequency points The corresponding activation confirmation duration is used to avoid accidental activation of the damping module due to short-term disturbances, sampling glitches, or transient impacts. When Φ s ( , k ) or Φ p ( , k The threshold for input judgment is reached, and at least two of the corresponding resonance criteria are continuously satisfied. T on Only then did the control unit confirm the existence of strong resonance characteristics at the candidate frequency point. T on If any of the necessary conditions are no longer met, the timer will restart.

[0082] Preferably, T on By candidate frequency points The period setting covers 4–12 resonant periods; within the 500–2000Hz range, this corresponds to approximately 2–24ms. It can also be set to satisfy the condition for 2–5 consecutive sliding time windows.

[0083] Similarly, when Φ p ( , k ) ≥ 1 and at least two of the parallel resonance-related criteria are simultaneously satisfied and persist. T on When 0.70 ≤ Φ, it is judged to have strong parallel resonance characteristics; p ( , k When ) < 1, it can be judged that the parallel resonance characteristics are weak or that it is in a warning state.

[0084] The above "stronger / weaker" judgment is not just a result description, but is directly used for subsequent control: the weaker state is used for early warning, increasing the sampling frequency or maintaining monitoring; the stronger state is used to trigger the selective activation of the corresponding damping branch and determine which type of damping module and which partition module to activate first.

[0085] Furthermore, the logic for damping input, tiered addition, and cutoff is as follows: Figure 1 The three sets of damping modules can be understood as multiple switchable damping modules located at different partition nodes or with different parameters. The control unit does not require them to be engaged simultaneously, but rather to selectively operate based on risk indicators and resonance type.

[0086] The preferred control logic is as follows: 1) When the series resonance risk index Φ s ( , k In candidate frequency points When the conditions for deployment are met, priority will be given to deploying devices that match the frequency point and are located within the regional circulation factor. K C Maximum or current amplification factor K I The RC damping branch and / or series RLC absorber branch at the largest partition node.

[0087] 2) When the parallel resonance risk index Φ p ( , k In candidate frequency points When the conditions for activation are met, priority should be given to activating the node voltage amplification factor. KU Maximum or peak impedance factor The series RLC absorber branch and / or low-pass damping branch at the largest node.

[0088] If the first damping module is put into operation, after the evaluation period... T eval The reduction ratio of the corresponding resonance risk index triggered by the damping module is less than the preset ratio. or drop For example, if the decrease is less than 20%, a second-level additional investment is executed. This second-level additional investment involves adding a second damping module while keeping the first damping module active. The second damping module is preferably selected from damping modules that match the same candidate frequency, the same resonance type, the same sensitive node, or the same sensitive zone. When adjacent zones simultaneously exhibit significant risks of zone circulating current, current amplification, node voltage amplification, or coupling, the second damping module is selected from damping modules of the adjacent zone.

[0089] If high risk occurs simultaneously at multiple candidate frequency points, or Φ s ( , k ), Φ p ( , k If any of the indicators exceeds the severe threshold Φcrit, then the combined deployment of multiple damping modules can be executed; this is only permitted under such severe or multi-frequency operating conditions. Figure 1 All or most of the damping modules are put into operation simultaneously.

[0090] The damping module is deactivated using hysteresis logic. The activation threshold is denoted as Φin, and the deactivation threshold as Φout, satisfying Φout < Φin. Preferably, Φin can be 1.00, and Φout can be 0.70. When the damping module is activated, the corresponding resonance risk index Φ... s ( , k ) or Φ p ( , kThe resection confirmation duration is lower than Φout and continues for the preset duration. T off Then, the control unit issues a resection command. The preset resection confirmation duration... T off Its function is to prevent the damping module from being immediately cut off when the risk indicator just falls below the release threshold, which would cause the damping module to be repeatedly engaged and disengaged near the threshold.

[0091] By using the hysteresis and delay control described above, frequent switching of each damping module branch near the threshold can be avoided.

[0092] Step 4) Estimating the ringing period, resonant frequency, and equivalent parameters is used to verify, supplement, and correct the frequency sweep identification results in Step 1), and to provide a basis for damping module parameter tuning. This step is not a necessary step for the control unit to switch the damping module online, but rather an offline identification and tuning step before commissioning or during maintenance and debugging.

[0093] Based on the offline frequency sweep identification in step 1), this invention further employs ringing testing as a supplementary identification method. Specifically, before the resonant suppression system is put into operation or during maintenance and commissioning, step excitation or pulse excitation is applied to the parallel capacitor bank and busbar network, and their damped oscillation response is observed. The control unit estimates the ringing frequency based on the interval between adjacent peaks in the same direction in the ringing waveform, and estimates the damping ratio and equivalent natural frequency based on the attenuation relationship of adjacent peak amplitudes. The ringing test results are used to verify, supplement, or correct the series resonant candidate frequency point set obtained by offline frequency sweep. F s Candidate frequency set of parallel resonance F p This serves as the basis for subsequent damping module parameter tuning.

[0094] By detecting adjacent peak times t1, t2, ..., tN in the same direction of the ringing waveform, the average ringing period Tr can be expressed as:

[0095] in, t r For the first r The time corresponding to each peak in the same direction; t r+1 For the first r +1 time corresponding to the same-direction peak; r Peak number; N This represents the total number of peak values ​​in the same direction that are included in the calculation.

[0096] Ringing frequency Estimated by the following formula:

[0097] If further estimation of the damping level is required, the amplitude of adjacent peak values ​​can be compared. A r and A r+1 Perform logarithmic decreasing calculation:

[0098] in: d It is a logarithmic decrease; By logarithmic decrease d Estimate the damping ratio ζ:

[0099] Under underdamped conditions, the ringing frequency can be further determined. Estimating the equivalent natural frequency f n : When the damping is relatively small and They are approximately equal. This explains "how to estimate the resonant frequency based on the ringing period": first, obtain the average period from the time interval between adjacent peak values ​​or zero-crossing points in the same direction, and then take its reciprocal to obtain the ringing frequency; if the damping effect is considered, then correct according to the above formula.

[0100] In a further embodiment, a known small capacitor may be added to the system. C x Then perform the ringing test again. If the previous frequency is... The frequency after addition is Then the equivalent capacitance and equivalent inductance can be derived:

[0101]

[0102] in, C eq The equivalent capacitance of the network to be identified; L eq The equivalent inductance of the network to be identified.

[0103] This provides a physical basis for the design of subsequent damping branch parameters, rather than relying on trial and error based on experience.

[0104] Step 5) Damping parameter design, integrated objective function and retest closed loop This step is used to illustrate the relationship between offline identification results and the parameter design of the damping module.

[0105] First, based on the candidate frequency points identified by frequency scanning... The equivalent natural frequency obtained from the ringing test Constructing and confirming target frequency When the relative deviation between the two does not exceed a preset threshold When this happens, the weighted result can be used as the confirmed target frequency:

[0106] in, The preferred values ​​are 3%–8%, with λ being a weight between 0 and 1. If the difference between the two is significant, both can be retained as candidate targets, and damping assessments can be performed separately.

[0107] 1. Design of branch parameters for a series RLC absorber. The preferred tuning frequency for the series RLC absorber is... near f *:

[0108] in, L d For damping inductance; C d It is a damping capacitor.

[0109] In preselect C d In this case, it can be obtained from the following formula. L d :

[0110] Damping resistor R d Based on the target damping ratio g d Or target quality factor Q d Conduct preliminary selection:

[0111] 2. RC series damping branch parameter design. Optimize its corner frequency. Located at the confirmed target frequency nearby:

[0112] 3. Low-pass damping branch parameter design. Assume... The low-frequency components that need to be bypassed as much as possible represent the frequency. To confirm the target frequency, the shunt reactor... L b It can be satisfied:

[0113] Where α is the low-frequency bypass constraint coefficient, preferably 0.1–0.3, and β is the target frequency band energy consumption constraint coefficient, preferably 0.8–1.5, thereby enabling the reactor to effectively bypass the damping resistor at low frequencies, while confirming the target frequency. Reduce bypass effect.

[0114] 4. Comprehensive Objective Function. This application proposes a comprehensive objective function for damping parameter tuning to balance between "resonance suppression effect" and "additional loss / current burden":

[0115] in, J The overall objective function; Z peak,old The peak impedance before damping is applied; Z peak,new This represents the peak impedance after the damper is applied. Z valley,old This represents the impedance valley value before the damper is applied. Z valley,new This represents the impedance valley value after the damper is applied. P d For the power consumption of the damping branch; P lim This refers to the allowable power consumption limit for the damping branch; I d For damping branch current; I lim λ1, λ2, λ3, and λ4 are the allowable current limits for the damped branch; λ1, λ2, λ3, and λ4 are weighting coefficients.

[0116] in, Used to evaluate the degree of decrease in peak impedance of parallel resonant circuits. Used to evaluate the degree of rise in the valley value of series resonant impedance. Used to limit power consumption in damped branches The damping parameters are used to limit the current stress in the damping branch, with λ1 to λ4 being weighting coefficients. By selecting the option with a smaller J from the candidate parameter set, damping parameters that balance vibration suppression effect and engineering feasibility can be obtained.

[0117] It should be noted that the basic calculations based on the equivalent relationship of RLC and RC belong to the conventional physical relationship of the circuit; while the comprehensive evaluation method of "incorporating the reduction of impedance peak value, the increase of impedance valley value, damping power consumption and branch current margin into the same setting target" is the design formula proposed in this application for the resonance suppression setting process of parallel large-capacity capacitor bank.

[0118] 5. Multiple rounds of retesting. After the initial parameters are determined, perform frequency sweep tests and / or ringing tests again, and compare the peak impedance Z before and after damping is applied. peak Impedance valley value Zvalley Current amplification factor K I , zonal circulation factor K C Node voltage amplification factor K U and Φ s ( , k ), Φ p ( , k If the target frequency is confirmed. Peak impedance Z peak No significant reduction, impedance valley Z valley No significant increase or regional circulation factor K C If the damping is not sufficiently reduced, adjust the damping module parameters, placement, or deployment strategy until the design objectives are met.

[0119] The entire system operation process is as follows: Before the resonant suppression system is put into operation or during maintenance and commissioning, firstly, according to step 1), perform offline frequency sweep identification on the target frequency band of 500–2000Hz to obtain the set of candidate frequency points for series resonance. F s Parallel Resonance Candidate Frequency Set F p .

[0120] When further verification of candidate frequencies, estimation of damping levels, or inverse calculation of equivalent parameters is required, a ringing test is performed according to step 4), and the ringing frequency, damping ratio, and equivalent natural frequency are obtained based on the damped oscillation response. Step 4) is a supplementary identification step to step 1) and is not a necessary step in the online operation phase; if step 4) is not performed, the subsequent damping parameter design is directly based on the candidate frequencies obtained in step 1).

[0121] Following step 5), the initial parameter tuning of each damping branch is completed, and the changes in impedance peak value, impedance valley value, zone circulating current, node voltage, and risk indicators before and after damping are verified by frequency sweep or ringing tests. The tuned damping module parameters, candidate frequency points, activation threshold, release threshold, and duration parameters are stored in the control unit.

[0122] Once the system enters online operation, the damping module is in the off state by default, and the control unit continuously collects signals such as bus voltage, total current, zone current, zone node voltage, damping branch current, damping module temperature, and switch status.

[0123] During the online operation phase, the control unit performs online preprocessing of the sampled signal according to step 2), constraining the candidate frequency points, around the series resonant candidate frequency point set. Fs Candidate frequency set of parallel resonance F p Candidate frequency points f q Extract the corresponding voltage amplitude, current amplitude, zone current amplitude, node voltage amplitude, and bandpass energy characteristics.

[0124] Subsequently, the control unit calculates the candidate frequency points according to step 3). In the current sliding time window k Internal series resonance risk index Φ s ( , k ) and parallel resonance risk index Φ p ( , k ), and determine the resonance type and sensitive zone based on the composite resonance risk index and quantitative criteria.

[0125] When candidate frequency points Series resonance risk index Φ at the location s ( , k The series resonance input threshold Φin is reached, and the input confirmation time is continuously met. T on When the control unit determines that the risk of series resonance at the candidate frequency point is increased, it will prioritize the use of RC damping branches or series RLC absorber branches that match the candidate frequency point, series resonance type and sensitive zone.

[0126] When candidate frequency points Parallel resonance risk index Φ at the location p ( , k The parallel resonance input threshold Φin is reached, and the input confirmation time is continuously met. T on When the control unit determines that the risk of parallel resonance at the candidate frequency point is increased, it prioritizes the use of series RLC absorber branches or low-pass damping branches that match the candidate frequency point, parallel resonance type, and sensitive node.

[0127] After the damping module completes the first-level activation, the control unit continues to update the candidate frequency points according to steps 2) and 3). The characteristic quantities and risk indicators at the location. If the evaluation period has elapsed... T eval Subsequently, the reduction rate of the corresponding resonance risk indicator triggered by the damping module's activation was less than the preset rate. or dropThen, a second-level additional input is executed, which, while keeping the first damping module connected, adds a second damping module that matches the same candidate frequency point, the same resonance type, the same sensitive node, or the same sensitive zone.

[0128] When the damping module is put into operation due to the risk of series resonance, if the candidate frequency point Series resonance risk index Φ at the location s ( , k Below the release threshold Φ out and continue to reach the confirmed resection duration. T off The control unit disconnects the corresponding damping module.

[0129] When the damping module is put into operation due to the risk of parallel resonance, if the candidate frequency point Parallel resonance risk index Φ at the location p ( , k The value is below the release threshold Φout and continues to reach the resection confirmation duration. T off The control unit disconnects the corresponding damping module.

[0130] Through the above methods, the system forms a closed-loop operation logic of "offline identification - parameter tuning - online detection - risk assessment - damping switching - effect verification - delayed release".

[0131] In addition to the external damping module, this invention also incorporates structured distributed damping as an additional vibration suppression structure. The structured distributed damping is arranged in the resonant sensitive zone nodes, capacitor terminal connection pieces, transition pieces, or local busbars, as identified offline, and forms finite damping by using a conductive material with a resistivity higher than copper. The structured distributed damping, together with the external damping module, reduces the sharpness of high-Q resonances and is only arranged in the resonant sensitive zones or nodes, avoiding a global increase in main power loss.

Claims

1. A system for suppressing resonance of a parallel large-capacity capacitor bank, characterized in that, include: Parallel capacitor bank, busbar system, zoned secondary busbar structure, damping module, sampling unit, control unit, and damping switching drive unit; The parallel capacitor bank includes at least two capacitor sections, each connected to the positive and negative busbars via a section busbar; the damping module includes multiple damping modules located at different section nodes and / or key busbar nodes, each connected to the corresponding node via an independent switching device; the sampling unit is used to collect bus voltage, total current, at least a portion of section current and / or section node voltage; the control unit pre-stores the series resonance candidate frequency set and the parallel resonance candidate frequency set obtained from offline identification, and performs online preprocessing and resonance feature extraction on the candidate frequency points to form series resonance risk indicators and parallel resonance risk indicators; When the corresponding risk indicator exceeds the input threshold and continues to meet the preset time, the damping switching drive unit controls at least one damping module corresponding to the resonance type, candidate frequency point and sensitive partition to be connected; when the corresponding risk indicator is lower than the release threshold and continues to meet the preset time, the unit controls at least one damping module to be disconnected.

2. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 1, characterized in that, The damping module includes at least one of the following: an RC series damping branch, a series RLC absorber branch, and a low-pass damping branch; the low-pass damping branch includes a parallel unit formed by a damping resistor and a parallel reactor, and the parallel unit is connected in series with a damping capacitor and then connected to the corresponding partition node.

3. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 1 or 2, characterized in that, The damping modules are not all activated simultaneously under normal conditions, but are selectively activated in at least one of the following ways: activation of a single damping module, tiered activation of multiple damping modules at the same frequency, coordinated activation of damping modules in adjacent zones, and combined activation under multi-frequency events; only when multiple candidate frequency points are triggered simultaneously or the resonance risk index reaches a severe threshold are all or most of the damping modules allowed to be activated.

4. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 1, characterized in that, The control unit performs preprocessing for candidate frequency point constraints on each candidate frequency point. The preprocessing includes at least: removing the DC component, establishing an adaptive narrowband window around the candidate frequency point, calculating the bandpass energy within the sliding time window, and using the Goertzel algorithm, single-frequency discrete Fourier analysis, or equivalent single-frequency amplitude estimation algorithm to obtain the voltage amplitude, current amplitude, and energy characteristics of the corresponding candidate frequency point.

5. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 1, characterized in that, The series resonance risk index is composed of at least the current amplification factor, impedance valley factor, zoned circulating current factor, and energy growth rate factor after weighted normalization; the parallel resonance risk index is composed of at least the node voltage amplification factor, impedance peak factor, disturbance sensitivity factor, and energy growth rate factor after weighted normalization.

6. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 5, characterized in that, When the control unit determines that the series resonance characteristics are strong, it prioritizes activating the RC damping branch and / or series RLC absorber branch located at the node with the largest zonal circulating current factor or the largest current amplification factor; when it determines that the parallel resonance characteristics are strong, it prioritizes activating the series RLC absorber branch and / or low-pass damping branch located at the node with the largest node voltage amplification factor or the largest impedance peak factor. This is done when the series resonance risk index is ≥ 1 and at least two of the series resonance criteria are simultaneously met and remain in effect for a certain period. T on When the series resonance risk index is 0.70 ≤ 1, it is judged as having strong series resonance characteristics. When the series resonance risk index is ≥ 1 and at least two of the parallel resonance criteria are met simultaneously and continuously, it is judged as having weak series resonance characteristics or being in a warning state. T on At that time, it was determined that the parallel resonance characteristics were strong; When 0.70 ≤ Parallel resonance risk index < 1, it is judged that the parallel resonance characteristics are weak or that it is in a warning state.

7. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 1, characterized in that, The series resonant candidate frequency set is obtained by the occurrence of local impedance valleys and the accompanying current amplitude amplification during the offline frequency sweep process; the parallel resonant candidate frequency set is obtained by the occurrence of local impedance peaks and the accompanying increase in node voltage response during the offline frequency sweep process.

8. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 7, characterized in that, The offline identification also includes ringing tests. Before the system is put into operation or during maintenance and debugging, step excitation or pulse excitation is applied to the parallel capacitor bank and busbar network, and their damped oscillation response is observed. The control unit estimates the ringing frequency based on the interval between adjacent peaks in the ringing waveform, and estimates the damping ratio and equivalent natural frequency based on the attenuation relationship of adjacent peak amplitudes. The ringing test results are used to verify, supplement or correct the parallel resonance candidate frequency point set and series resonance candidate frequency point set obtained by offline frequency sweep.

9. The system for suppressing resonance of a parallel large-capacity capacitor bank according to claim 1, characterized in that, The parameters of the damping module are tuned based on the confirmed target frequency identified offline: the resonant frequency of the series RLC absorber branch is close to the confirmed target frequency; the cutoff frequency of the RC series damping branch is near the confirmed target frequency; and the parameters of the parallel reactor in the low-pass damping branch satisfy both low-frequency bypass and the confirmed target frequency. The dual constraints of energy consumption; The damping module parameter tuning is also based on a comprehensive objective function that takes into account both resonance suppression effect and additional losses. The comprehensive objective function includes at least two or more of the following: impedance peak reduction ratio, impedance valley rise ratio, damping branch power consumption ratio, and damping branch current margin.

10. A method for suppressing resonance in a parallel large-capacity capacitor bank system according to any one of claims 1-9, characterized in that, Includes the following steps: S1, offline frequency sweep identification and / or ringing test are used to obtain the candidate frequency points of series resonance and the candidate frequency points of parallel resonance; S2, online acquisition of bus voltage, total current, zone current and / or zone node voltage; S3 performs DC removal, adaptive narrowband filtering, sliding window energy calculation, and single-frequency amplitude estimation around the candidate frequency point; S4, construct series resonance risk indicators and parallel resonance risk indicators; S5. Selectively deploy at least one damping module based on the magnitude, duration and location of the risk indicator and the sensitive zone. S6, after the risk indicator is lower than the release threshold and continues to meet the preset time, the corresponding damping module is cut off; S7, based on the changes in impedance peak and valley, branch current and node voltage before and after the damping is put into operation, the damping parameters are re-measured and adjusted.

Images