A shock-resistant damping method for a transformer with elastic damping elements

Abstract

The application relates to the technical field of transformer vibration control, and particularly discloses a transformer anti-impact damping method with an elastic damping element, which obtains transformer related parameters, extracts a transient electrical characteristic vector through wavelet transient characteristic extraction, and filters an acceleration signal to obtain a net mechanical vibration state matrix in combination with historical modal operation parameters; an impact working condition trigger identifier is generated through energy out-of-limit determination of the transient electrical characteristic vector, and when the identifier is valid, cross-domain feedforward mechanical prediction solving is performed on the identifier to obtain a Lorentz force peak value estimation matrix and a stress wave arrival time difference; the identifier is used for double-modal switching and synthesis of an excitation current target matrix, and stiffness and damping reconstruction is completed through power amplification injection into the excitation coil of the magneto-rheological elastic damping element. Since electrical calculation is much faster than stress wave conduction, the element can complete preset hardening before a destructive impact arrives, get rid of a posteriori feedback time lag, and also avoid misoperation caused by mechanical sensing failure under strong magnetic leakage.

Description

Technical Field

[0001] This application relates to the field of transformer vibration control technology, and more specifically, to a method for shock-resistant vibration reduction of transformers with elastic damping elements. Background Technology

[0002] Large power transformers are constantly subjected to two types of mechanical excitations during operation: one is the steady-state high-frequency vibration of the silicon steel sheets in the core due to the magnetostrictive effect under an alternating magnetic field; the other is the transient low-frequency impact generated by the Lorentz force produced by the interaction of winding currents during sudden load changes or external short circuits. The latter can reach hundreds of kilonewtons within milliseconds, which is enough to cause winding deformation, loosening of clamps, and even rigid collision between the transformer body and the tank, as well as insulation damage.

[0003] Existing transformer vibration reduction methods with elastic damping elements are mostly based on mechanical closed-loop systems: first, the vibration signal is collected by the acceleration or displacement sensor installed in the oil tank, and then the excitation current of the magnetorheological damper or magnetorheological elastomer is adjusted through filtering and control calculations, thereby changing the damping and stiffness; passive rubber pads and multilayer elastomer vibration isolation devices rely entirely on fixed stiffness and damping parameters. The common flaw of this type of solution lies in the fact that the control closed loop is based on a posterior causal link of mechanical force generation, vibration occurrence, sensor sampling, and damper adjustment: it takes more than ten milliseconds for the stress wave to be transmitted from the winding to the damping element, several milliseconds for sensor sampling, filtering, and control calculation, and more than ten to tens of milliseconds for the excitation coil to complete magnetization under conventional constant voltage source drive mode without strong excitation measures. The cumulative delay of the above three segments is much greater than the arrival time of the first peak of the short-circuit electromagnetic force, which means that the most destructive first wave of impact energy cannot be isolated at all. At the same time, the spatial leakage magnetic field increases sharply at the moment of short circuit, the mechanical sensor signal is severely interfered with at the moment when it is most needed to play its role, and the feedback link may output incorrect instructions, making the vibration reduction and shock resistance defense line ineffective.

[0004] Furthermore, even if the condition determination is introduced to achieve mode switching between feedforward and feedback, the binarized impact indicator can only represent whether the impact determination condition is met. It cannot reflect the decay state of the residual mechanical energy in the structure after the impact is removed. The steady-state feedback channel is prone to mistreating the low-frequency amplitude residual vibration that has not yet been dissipated as a steady-state error, which in turn causes integral saturation, differential spikes and damping stiffness flutter, prolonging the transition time from the impact condition to the steady-state vibration reduction condition.

[0005] Therefore, an optimized shock and vibration reduction scheme for transformers with elastic damping elements is desired. Summary of the Invention

[0006] To address the aforementioned technical problems, this application is proposed. Embodiments of this application provide a method for shock resistance and vibration reduction of a transformer with an elastic damping element, comprising: S1, acquire the three-phase transient current signal on the primary side of the transformer, the mechanical acceleration signal of the transformer body tank, and historical mode operating parameters; S2, perform wavelet transient feature extraction on the three-phase transient current signal of the primary side of the transformer to obtain the transient electrical feature vector, and perform Kalman filtering on the tank mechanical acceleration signal based on the historical mode operating parameters to obtain the net mechanical vibration state matrix; S3, perform energy limit determination on transient electrical feature vectors to obtain impact condition triggering identifier; S4, in response to the effective setting of the impact condition triggering flag, performs cross-domain feedforward mechanical prediction calculation on the transient electrical characteristic vector to obtain the Lorentz force peak prediction matrix and stress wave arrival time difference; S5, based on the impact condition triggering flag, performs dual-mode switching to obtain the solution result of the current effective mode, and synthesizes the solution result of the current effective mode into the excitation current target matrix; S6, after power amplification, injects the target matrix of excitation current into the excitation coil of the magnetorheological elastic damping element to reconstruct stiffness and damping.

[0007] Compared with existing technologies, this application proposes a method for shock-resistant vibration reduction of transformers with elastic damping elements. It simultaneously acquires the three-phase transient current signal of the transformer primary side, the mechanical acceleration signal of the transformer tank, and historical mode operating parameters. Wavelet transient feature extraction is performed on the three-phase transient current signal to obtain transient electrical feature vectors. Kalman filtering is applied to the acceleration signal using historical mode operating parameters to obtain the net mechanical vibration state matrix. An impact condition trigger flag is generated by determining the energy limit of the transient electrical feature vectors. When the flag is validly set, cross-domain feedforward mechanical prediction calculation is performed on the transient electrical feature vectors to directly derive the Lorentz force peak value prediction matrix and the stress wave arrival time difference. Then, using this flag, dual-mode switching is performed, and the calculation results of the current effective mode are synthesized into an excitation current target matrix. This matrix is ​​then amplified and injected into the excitation coil of the magnetorheological elastic damping element to complete stiffness and damping reconstruction. Since the propagation and calculation of electrical signals are much faster than the propagation of mechanical stress waves, the scheme can complete the component hardening and energy consumption state presetting before the arrival of destructive stress waves, without waiting for mechanical vibration to be measured. This eliminates the inherent time delay of the a posteriori feedback link, avoids the risk of malfunction caused by the failure of mechanical sensing channels under strong leakage magnetic interference, and returns to steady-state micro-vibration suppression after the impact dissipates. Attached Figure Description

[0008] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 This is a flowchart of a method for shock resistance and vibration reduction of a transformer with an elastic damping element according to an embodiment of this application; Figure 2 This is a data flow diagram illustrating a method for shock resistance and vibration reduction of a transformer with an elastic damping element according to an embodiment of this application. Figure 3 This is a flowchart illustrating the process of determining the energy limit of transient electrical characteristic vectors to obtain the shock condition triggering identifier in a method for shock-resistant vibration reduction of a transformer with elastic damping elements according to an embodiment of this application. Figure 4 In response to the effective setting of the shock condition triggering flag, a flowchart is used to perform cross-domain feedforward mechanical prediction calculation on the transient electrical characteristic vector to obtain the Lorentz force peak value prediction matrix and the stress wave arrival time difference in a transformer shock damping method according to an embodiment of this application. Figure 5 This is a flowchart illustrating a method for shock resistance and vibration reduction of a transformer with elastic damping elements according to an embodiment of this application. The method involves performing dual-mode switching based on a shock condition triggering identifier to obtain the solution result of the current effective mode, and then synthesizing the solution result of the current effective mode into an excitation current target matrix. Figure 6 This is a flowchart illustrating the process of modal separation of the net mechanical vibration state matrix and calculation based on the difference between the steady-state harmonic channel and the impact residual channel to obtain the feedback steady-state adjustment target current component in a method for shock-resistant vibration reduction of a transformer with elastic damping elements according to an embodiment of this application. Detailed Implementation

[0010] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0011] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not specifically singular and may include plural forms. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0012] Flowcharts are used in this application to illustrate the operations performed by the system according to embodiments of this application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, various steps can be processed in reverse order or simultaneously as needed. Furthermore, other operations can be added to these processes, or one or more steps can be removed from them.

[0013] Figure 1 This is a flowchart of a method for shock and vibration reduction of a transformer with an elastic damping element according to an embodiment of this application. Figure 2 This is a data flow diagram illustrating a method for shock resistance and vibration reduction of a transformer with an elastic damping element according to an embodiment of this application. Figure 1 and Figure 2 As shown, an embodiment of this application provides a method for shock and vibration reduction of a transformer with elastic damping elements, comprising: S1, acquiring the three-phase transient current signal on the primary side of the transformer, the mechanical acceleration signal of the transformer tank, and historical mode operating parameters; S2, performing wavelet transient feature extraction on the three-phase transient current signal on the primary side of the transformer to obtain a transient electrical feature vector, and performing Kalman filtering on the tank mechanical acceleration signal based on the historical mode operating parameters to obtain a net mechanical vibration state matrix; S3, performing energy limit judgment on the transient electrical feature vector. S4. In response to the effective setting of the impact condition triggering flag, cross-domain feedforward mechanical prediction calculation is performed on the transient electrical characteristic vector to obtain the Lorentz force peak value prediction matrix and stress wave arrival time difference; S5. Based on the impact condition triggering flag, dual-mode switching is performed to obtain the calculation result of the current effective mode, and the calculation result of the current effective mode is synthesized into the excitation current target matrix; S6. The excitation current target matrix is ​​amplified by power and injected into the excitation coil of the magnetorheological elastic damping element for stiffness and damping reconstruction.

[0014] Specifically, in step S1, the three-phase transient current signal of the transformer primary side, the mechanical acceleration signal of the transformer tank, and historical modal operating parameters are acquired. It should be noted that since the transformer's shock resistance and vibration reduction control relies on both electrical quantities that can anticipate short-circuit transients and mechanical quantities that characterize the actual vibration state of the transformer body, data from a single source is insufficient to support the simultaneous determination of shock conditions and dual-modal control. Therefore, the technical solution of this application first acquires the three-phase transient current signal of the transformer primary side, the mechanical acceleration signal of the transformer tank, and historical modal operating parameters. Through the above processing, it is possible to effectively provide raw data with a clear source and consistent time base for subsequent transient characteristic calculation and mechanical state estimation.

[0015] More specifically, in a particular example of this application, the three-phase transient current signal on the primary side of the transformer is acquired by a current transformer mounted on the three-phase leads of the primary side and a subsequent high-bandwidth conditioning circuit, with a sampling rate of not less than 20,000 times per second to ensure that the steepness information of the short-circuit current rising edge is not lost due to undersampling. This signal is used to characterize the abrupt change in the amplitude and rate of change of the primary side current when an external short circuit occurs. The mechanical acceleration signal of the transformer tank is acquired by a piezoelectric accelerometer that is adhesively or magnetically mounted on the tank wall and bottom support, with the measurement direction... Covering both horizontal and vertical directions, the measurement range encompasses the power frequency harmonic micro-vibrations caused by core magnetostriction and the low-frequency amplitude vibrations caused by short-circuit impacts. This signal characterizes the actual vibration response at the installation location of the elastic damping element. Historical modal operating parameters are read from the structure in the dual-port random access memory, which is written back from the previous control cycle. This includes the process noise covariance obtained from the physical excitation current tracking error matrix and the current control mode duration. During the initial control cycle upon power-on, these parameters are set to the factory-calibrated default constant. The acquisition of the above three data streams is initiated at the beginning of the same control cycle. The analog-to-digital conversion of the current and acceleration signals is uniformly triggered by the global hardware clock of the field-programmable gate array, aligning their sampling times to the microsecond level to avoid timing ambiguity in subsequent impact condition determination due to sampling offsets between the electrical and mechanical channels. The historical modal operating parameters are read in as unpacked data and sent to the subsequent feature extraction and filtering stages along with the two real-time signals.

[0016] Specifically, in step S2, wavelet transient feature extraction is performed on the three-phase transient current signal on the primary side of the transformer to obtain a transient electrical feature vector, and Kalman filtering is performed on the tank mechanical acceleration signal based on historical mode operating parameters to obtain a net mechanical vibration state matrix. It should be noted that the short-circuit characteristics in the original three-phase transient current signal are masked by the power frequency fundamental wave, and the tank mechanical acceleration signal contains environmental noise and sensor measurement noise, neither of which can be directly used as a control basis. Therefore, the technical solution of this application further performs wavelet transient feature extraction on the three-phase transient current signal on the primary side of the transformer to obtain a transient electrical feature vector, and Kalman filtering is performed on the tank mechanical acceleration signal based on historical mode operating parameters to obtain a net mechanical vibration state matrix. Through the above processing, transient features characterizing short-circuit impacts can be effectively extracted on the electrical side, and a reliable vibration state description can be obtained on the mechanical side.

[0017] Specifically, in this embodiment, step S2 includes: synchronously acquiring the three-phase transient current signal and the tank mechanical acceleration signal through analog-to-digital conversion based on a global hardware clock to obtain a discretized three-phase current sequence and a discretized acceleration sequence; unpacking the historical mode operating parameters to extract the process noise covariance; performing wavelet domain electrical transient feature calculation on the discretized three-phase current sequence to obtain a transient electrical feature vector; initializing the Kalman filter with prior covariance weights based on the process noise covariance; and performing state recursion and measurement update filtering on the discretized acceleration sequence to obtain a net mechanical vibration state matrix.

[0018] More specifically, in a specific example of this application, the global hardware clock of the field-programmable gate array is used as a unified time base to initiate synchronous analog-to-digital conversion acquisition of the three-phase transient current signal on the primary side of the transformer and the mechanical acceleration signal of the transformer tank. The start pulses for the two conversions come from the frequency division output of the same counter, so that the sampling time deviation between the electrical channel and the mechanical channel is controlled within the microsecond range. The acquisition results are serialized into discrete three-phase current sequences and discrete acceleration sequences arranged by phase, respectively. At the same time, the historical mode operation parameters written back from the previous control cycle are unpacked, and the process noise covariance term is located and read out according to the memory layout of the structure. This covariance term quantifies the execution uncertainty exhibited by the constant current driver and the magnetorheological damping element in the previous cycle.

[0019] Based on the obtained discretized three-phase current sequence, wavelet domain electrical transient characteristic calculation is performed for each phase. That is, the Marat fast algorithm is independently applied to each phase subsequence of the sequence to perform discrete wavelet transform, decompose and remove the low-frequency approximation component of the power frequency fundamental wave layer by layer, and retain the high-frequency detail coefficients that characterize the abrupt change in the rising edge of the current of that phase. Then, the sum of squares of the detail coefficients of that layer is calculated within a sliding window to obtain the absolute value of the transient energy of that phase, the expression of which is:

[0020] in, Indicates the first The absolute value of the transient energy of the phase current quantum sequence within the current sliding window. This represents the phase sequence index, with each value corresponding to one of the three phases. This indicates the total number of discrete sampling points contained within the window. Indicates the index number of the discrete sampling point. Indicates the first After wavelet transform, the phase current subsequence is at the 1st The first layer at the decomposition scale The high-frequency detail coefficients at each sampling point are used to measure the degree of transient energy accumulation of each phase in the wavelet domain in the form of the sum of squares of the detail coefficients. When the transformer is running at power frequency in steady state, the values ​​of the three phases are maintained at low levels. When an external short circuit occurs on the primary side, the value of the fault phase rises with the sudden change of current. The rise of the non-fault phase is smaller than that of the fault phase. Thus, the phase-to-phase asymmetric information of the short circuit is preserved in the energy dimension.

[0021] Based on this, the slope abrupt change rate within a unit time window is calculated for each phase subsequence, and zero-crossing detection is performed on the low-frequency approximate component of the power frequency fundamental wave to extract the initial phase angle of each phase current. Subsequently, the absolute values ​​of the transient energy of the three phases, the slope abrupt change rate of the three phases, and the initial phase angle of each phase are concatenated according to their dimensions to form a transient electrical feature vector, the expression of which is:

[0022] in, This represents the transient electrical feature vector generated by splicing. , and These represent the absolute values ​​of the transient energies of the three phases, respectively. , and These represent the three-phase current subsequences at the sampling interval. The ratio of the current increment within the interval to that interval is used to characterize the steepness of the current rise time for each phase. This represents the closing phase angle at the moment of short circuit occurrence, obtained from the zero-crossing detection of the low-frequency approximate component. The aforementioned components collectively characterize the same electromagnetic transient process in the energy, rate of change, and phase dimensions. The phase-separated arrangement allows for the differentiation of different current distributions resulting from single-phase-to-ground short circuits, two-phase short circuits, and three-phase short circuits at the characteristic level. The closing phase angle component reflects the bias degree of the DC component in the short-circuit current. Together, they constitute the electrical input upon which subsequent impact condition determination and cross-domain mechanical prediction are based.

[0023] In parallel with the electrical side calculation, the process noise covariance obtained from the unpacking is written into the state-space equation of the Kalman filter. Based on this, the initial weights of the prior prediction covariance matrix are assigned, so that the filter's estimation of the execution link uncertainty matches the current actual driving state of the transformer. Subsequently, using the discretized acceleration sequence as the measurement input, state recursion and measurement updates are alternately executed according to the control cycle. The expression for the measurement update step is as follows:

[0024] in, Indicates the first The posterior state estimation vector updated by measurement for each control cycle. This represents the prior state prediction vector obtained recursively from the previous cycle state and the dynamic model of the fuel tank structure. Indicates the first The Kalman gain matrix for each control cycle is constrained by the aforementioned process noise covariance. Indicates the first The actual measurement values ​​read from the discretized acceleration sequence in each control cycle. This represents the measurement matrix that maps the state space to the observation space. In the formula... This is a novel term, representing the deviation between actual measurements and model predictions. It is weighted by Kalman gain to correct prior predictions, thereby suppressing background high-frequency environmental noise and sensor measurement white noise. The posterior state estimation vectors of each control cycle are arranged and combined in chronological order to obtain the net mechanical vibration state matrix. This matrix contains the displacement and velocity states at the installation positions of the elastic damping elements, and is used for calculating the feedback current under steady-state modes.

[0025] Specifically, in step S3, the transient electrical characteristic vector is subjected to energy limit exceedance determination to obtain the impact condition triggering identifier. It should be noted that, given that the current fluctuation of the transformer during steady-state operation differs from the current surge during an external short circuit by several orders of magnitude, if a discrimination limit is not set for the transient severity reflected by the transient electrical characteristic vector, the feedforward anti-surge path cannot determine when to intervene, and the two control requirements of steady-state micro-vibration suppression and impact advance defense cannot be distinguished. Moreover, the first peak of the short-circuit electromagnetic force arrives within ten milliseconds, and the discrimination step must give a conclusion within the current control cycle in which the electrical characteristic appears. Based on this, the technical solution of this application further performs energy limit exceedance determination on the transient electrical characteristic vector to obtain the impact condition triggering identifier, that is, maps the two components of transient energy absolute value and slope surge rate to the electrical surge energy absolute value characterizing the current electromagnetic transient severity, performs differential comparison with the threshold preset according to the winding insulation safety tolerance, and performs step logic binarization to obtain the impact condition triggering identifier that is set or cleared. Through the above processing, it is possible to effectively determine whether the impact condition is established before the stress wave reaches the elastic damping element, providing a unified criterion for the initiation of cross-domain feedforward mechanical prediction, the gating enable of nanosecond-level timing synchronization clock, and the weight allocation of feedforward and feedback paths, thereby avoiding the ineffective adjustment of excitation current caused by the false triggering of the anti-impact mode under normal operating conditions.

[0026] Figure 3 This is a flowchart illustrating the process of determining energy limits in transient electrical characteristic vectors to obtain an impact condition triggering indicator, according to an embodiment of this application, for a transformer with elastic damping elements designed for shock resistance and vibration reduction. Figure 3 As shown, step S3 includes: S31, performing electrical mutation energy spatial mapping on the transient electrical feature vector to obtain the absolute value of electrical mutation energy; S32, performing differential comparison between the absolute value of electrical mutation energy and the preset insulation safety tolerance threshold, and performing binarization judgment through step logic to obtain the impact condition triggering flag; S33, responding to the high-level setting of the impact condition triggering flag, enabling the gate-level oscillator inside the gate array to generate a timing synchronization control clock signal.

[0027] In step S31, the transient electrical feature vector is spatially mapped using electrical mutation energy to obtain the absolute value of the electrical mutation energy. It should be noted that since the absolute value of the transient energy in the transient electrical feature vector and the slope mutation rate have different physical dimensions and orders of magnitude, direct comparison with a single threshold cannot yield consistent discrimination results. Therefore, the technical solution of this application further performs spatial mapping of the transient electrical feature vector using electrical mutation energy to obtain the absolute value of the electrical mutation energy. Through the above processing, multi-dimensional electrical features can be effectively aggregated into a scalar that can be directly compared with the insulation safety tolerance threshold.

[0028] More specifically, in a specific example of this application, the transient electrical feature vector obtained from the previous calculation is read, and the absolute value component of transient energy and the slope change rate component are extracted one by one. Normalized weighting coefficients are then assigned to each component according to its range under the rated operating conditions of the transformer, ensuring that the wavelet domain energy term and the current rise steepness term are numerically of the same order of magnitude, thus preventing the large numerical range of the current change rate from dominating the aggregation result. Subsequently, a second-norm space mapping calculation is performed on each weighted component, i.e., the squares of each weighted component are first taken and summed, and then the square root of the summation is taken to obtain the absolute value of the electrical change energy characterizing the severity of the current electromagnetic transient, expressed as:

[0029] in, This represents the absolute value of the electrical mutation energy obtained through spatial mapping. This represents the total number of feature components contained in the transient electrical feature vector. The index number representing the characteristic component. The first eigenvector represents the transient electrical eigenvector. Each feature component Represented as the first The dimensionless normalized weight coefficients configured for each feature component. In the formula... The sum of squares of the weighted components after dimensionless scaling is the composite modulus of each characteristic component in the normalized characteristic space. This modulus remains low during the steady-state operation of the transformer at the power frequency and increases when an external short circuit occurs on the primary side and the current amplitude and rising edge steepness rise simultaneously. Thus, it reflects the transient over-limit degree of the electrical side in a single scalar form and serves as the direct input for subsequent differential comparison with the insulation safety tolerance threshold.

[0030] In step S32, the absolute value of the electrical transient energy is differentially compared with a preset insulation safety tolerance threshold, and a step logic is used for binarization to obtain the impact condition trigger indicator. It should be noted that since the absolute value of the electrical transient energy is a continuous scalar, it cannot directly drive the start-up of the feedforward path and the selection of the control mode; downstream processes require unambiguous logical criteria. Based on this, the technical solution of this application further differentially compares the absolute value of the electrical transient energy with the preset insulation safety tolerance threshold and uses step logic for binarization to obtain the impact condition trigger indicator. Through the above processing, the degree of limit violation on the electrical side can be effectively transformed into a binary criterion that can be used for cross-domain prediction and dual-mode switching within a single control cycle.

[0031] More specifically, in a specific example of this application, a preset insulation safety tolerance threshold is read from the constant area of ​​the memory. This threshold is obtained by inverting the short-circuit withstand capability of the transformer winding and the limit electrodynamic force allowed by the solid insulation components, and then calibrating it after being inverted from the short-circuit test data to the normalized characteristic space where the absolute value of the electrical transient energy is located. A margin is left on the calibrated value to cover the normal disturbances caused by load switching and inrush current. Then, the absolute value of the electrical transient energy of the current control cycle is subtracted from the threshold to obtain a difference result reflecting the direction and amplitude of the over-limit. This difference result is sent to the step logic for binarization conversion. That is, when the difference result is greater than zero, it is determined that an external short-circuit impact has occurred on the primary side, and the logic register where the impact condition trigger flag is located is set to 1. When the difference result is less than or equal to zero, it is determined that it is a normal steady-state operation or a load disturbance with limited amplitude, and the logic register is cleared to 0. The expression is as follows:

[0032] in, This indicates the impact condition trigger flag stored after binarization determination. This represents the absolute value of the electrical catastrophe energy obtained from the preorder space mapping. This represents the preset insulation safety tolerance threshold. The formula divides the continuous transient severity into two mutually exclusive intervals: impact condition and steady-state condition, with the threshold as the boundary. The determination result is stored in the logic register of the gate array in single-bit form. Within the same control cycle, it is distributed in parallel to the cross-domain feedforward mechanical prediction stage as a gating condition, to the enable terminal of the timing synchronization control clock as an activation signal, and to the dual-mode fusion stage as the weight source for the feedforward path and the feedback path. This ensures that the intervention time of the impact mode and the appearance time of the electrical characteristics are in the same control cycle, without introducing additional time delay due to the determination stage.

[0033] In step S33, in response to the high-level setting of the impact condition trigger flag, the nanosecond-level oscillator inside the gate array is gated and enabled to generate a timing synchronization control clock signal. It should be noted that since the feedforward calculation under the shock mode must be completed before the stress wave reaches the elastic damping element, the clock cycle of the conventional control period is insufficient to support the computational scheduling within this time window. Therefore, the technical solution of this application further responds to the high-level setting of the impact condition trigger flag by gated and enabling the nanosecond-level oscillator inside the gate array to generate a timing synchronization control clock signal. Through the above processing, a hard real-time clock can be effectively provided for the feedforward current calculation, and no unnecessary high-frequency switching is introduced during the steady-state period.

[0034] More specifically, in a specific example of this application, the output of the logic register storing the impact condition trigger flag is directly wired to the clock enable pin of the nanosecond-level oscillator inside the field-programmable gate array (FPGA). This allows the gating relationship to be established in combinational logic form without software polling. When the flag is set to 1 due to an external short circuit on the primary side, the enable pin is turned on at the next gate array reference clock edge. The oscillator outputs a continuous square wave pulse sequence, constituting a timing synchronization control clock signal with a pulse period on the order of nanoseconds. This signal drives the inverse calculation involved in the feedforward pre-bias current calculation to be completed within a millisecond-level stress wave propagation time window, and prioritizes the use of the gate array's computing resources and bus bandwidth for this operation. When the flag is cleared to 0, the enable pin is locked, the oscillator output is clamped to a constant zero potential, the timing synchronization control clock signal is disabled, and the feedforward calculation path remains silent, avoiding increased dynamic power consumption and electromagnetic emissions of the gate array due to continuous high-frequency clock flips during the transformer's steady-state operation at the power frequency. The generation relationship of this gating clock is expressed as follows:

[0035] in, Indicates at time The output timing synchronization control clock signal, This indicates the impact condition trigger flag, which serves as the gating enable coefficient; its value is either 1 or 0. This indicates the rated trigger amplitude voltage of the clock pulse. Indicates the period index number of the pulse sequence. This represents the fundamental oscillation period of a nanosecond-level oscillator. This indicates the duration of the high-level pulse of a single square wave pulse. This represents a rectangular window function. The summation term in the formula is calculated periodically with respect to the rectangular window. An ideal square wave pulse sequence is constructed by translation and superposition. This sequence is multiplied by the impact condition trigger flag, ensuring that the existence of the pulse sequence is entirely governed by the impact condition determination result. This binds the clock resource activation time with the short-circuit impact occurrence time at the hardware level. Furthermore, the timing synchronization control clock signal and the impact condition trigger flag are output together; the former serves as the activation condition for feedforward calculation, while the latter continues to be used for gating and bimodal weight allocation in cross-domain mechanics prediction.

[0036] Specifically, in step S4, in response to the effective setting of the impact condition trigger flag, a cross-domain feedforward mechanical prediction calculation is performed on the transient electrical characteristic vector to obtain the Lorentz force peak value prediction matrix and the stress wave arrival time difference. It should be noted that, given that the a posteriori feedback link based on the tank mechanical acceleration signal requires the stress wave to be transmitted from the winding to the elastic damping element, processed by sampling and filtering, and magnetized by the excitation coil before the damping and stiffness can be changed, its cumulative time delay is much greater than the arrival time of the first peak of the short-circuit electromagnetic force. Furthermore, the impact condition trigger flag itself only indicates whether a short circuit has occurred and does not carry information on the force amplitude and arrival time of the force that will soon act on the transformer body. Therefore, it is impossible to determine the target amount of the excitation current and the start and end sequence of magnetization based on this information. Based on this, the technical solution of this application further responds to the effective setting of the impact condition triggering flag by performing cross-domain feedforward mechanical prediction calculation on the transient electrical characteristic vector to obtain the Lorentz force peak value prediction matrix and the stress wave arrival time difference. That is, using this flag as a gating condition, the transient short-circuit current of the winding is reconstructed from the transient electrical characteristic vector. The peak distribution of electrodynamic force acting on each segment of the winding is derived through the spatial leakage magnetic coupling relationship, and the time delay of the cross-medium propagation path of the stress wave is extrapolated based on the wave velocity of the winding material and the sound velocity of the insulating oil. Through the above processing, two quantitative results, force amplitude and available time window, can be effectively obtained before the arrival of the destructive stress wave. This provides a clear basis for the stiffness hardening degree of the damping element and the rise rate of the magnetization, thereby shifting the control basis from the already occurred mechanical vibration to the faster propagating electrical quantity and avoiding false operation caused by the failure of the mechanical sensing channel under the strong leakage magnetic interference at the moment of short circuit.

[0037] Figure 4 This document describes a flowchart illustrating a method for shock-resistant vibration reduction of a transformer with elastic damping elements, based on an embodiment of this application. It describes a process for performing cross-domain feedforward mechanical prediction on transient electrical characteristic vectors to obtain the Lorentz force peak value prediction matrix and stress wave arrival time difference in response to the effective setting of an impact condition trigger flag. (The flowchart is not translated here.) Figure 4 As shown, step S4 includes: S41, using the impact condition triggering identifier as the gated multiplication coefficient, performing transient feature gated modulation and short-circuit current reconstruction on the transient electrical feature vector to obtain the gated effective short-circuit current vector; S42, based on the leakage magnetic coupling coefficient matrix and winding geometric topology parameters, performing Lorentz force calculation under spatial leakage magnetic coupling on the gated effective short-circuit current vector to obtain the Lorentz force peak prediction matrix; S43, based on the spatial coordinates of the force application point in the Lorentz force peak prediction matrix and the preset winding material wave velocity constant and insulating oil sound velocity constant, performing time delay extrapolation on the stress wave cross-medium propagation path to obtain the stress wave arrival time difference.

[0038] In step S41, the impulse condition trigger flag is used as the gated multiplication coefficient to perform transient characteristic gated modulation and short-circuit current reconstruction on the transient electrical characteristic vector to obtain the gated effective short-circuit current vector. It should be noted that the transient electrical characteristic vector carries the wavelet domain energy, current rise steepness, and closing phase angle of each phase, and is not a current quantity that can be directly used for electrodynamic calculations. Furthermore, it should not trigger the feedforward path calculation under steady-state conditions. Based on this, the technical solution of this application further utilizes the impulse condition trigger flag as the gated multiplication coefficient to perform transient characteristic gated modulation and short-circuit current reconstruction on the transient electrical characteristic vector to obtain the gated effective short-circuit current vector. Through the above processing, the phase-separated electrical characteristics can be effectively restored to the winding transient short-circuit current distribution, while the activation conditions of the feedforward path are constrained algebraically.

[0039] More specifically, in a specific example of this application, the impact condition trigger flag registered in the logic register is read, and its value of 1 or 0 is used as the gating coefficient of the multiplier and applied to the transient electrical characteristic vector. Algebraic mask modulation is performed on the vector so that when the flag is set, the vector passes through as is, and when the flag is cleared, all components of the vector are reduced to zero. This achieves gating of the feedforward path without adding conditional jump instructions, avoiding the introduction of additional scheduling delays within the millisecond time window due to software branch judgments. Subsequently, the modulated vector is multiplied by a preset characteristic current mapping weight matrix, and a linear combination is performed. The short-circuit current reconstruction uses a weight matrix whose input dimension matches the dimension of the transient electrical feature vector and whose output dimension corresponds to the three-phase winding. The coefficients in each row are obtained by offline fitting of the three-phase current waveforms recorded during the transformer's factory short-circuit test with their corresponding phase wavelet domain energy, phase slope abrupt change rate, and closing phase angle samples. Different coefficient segments are defined for single-phase ground faults, two-phase short circuits, and three-phase short circuits, allowing the reconstruction process to locate the fault type segment based on the relative magnitude of each phase's characteristics and then calculate the current peak value phase by phase. The reconstruction result constitutes a gated effective short-circuit current vector, the expression of which is:

[0040] in, This represents the gated effective short-circuit current vector obtained from the reconstruction, whose three components correspond to the peak transient short-circuit current of each of the three phase windings. This indicates the triggering identifier for the impact condition, which serves as the gating multiplication coefficient. This represents the preset characteristic current mapping weight matrix, whose number of rows equals the number of three-phase windings and whose number of columns equals the dimension of the transient electrical eigenvectors. This represents the transient electrical characteristic vector. In the formula... The mapping result, which returns the current dimension from the normalized feature space, avoids the component linear correlation problem caused by insufficient dimension in solving the three-phase current peak value problem because the input dimension is higher than the output dimension. The mapping result can reflect the phase-to-phase asymmetry of the short circuit. After multiplying this result with the impact condition trigger flag, the true three-phase current peak value distribution is output when an external short circuit occurs on the primary side, and the zero vector is output when the transformer is running at power frequency in steady state, so that the subsequent Lorentz force calculation does not produce a non-zero force prediction during steady state. In addition, the impact condition trigger flag is used as a gating coefficient in this step and continues to be passed down for subsequent time delay derivation and dual-mode weight allocation.

[0041] In step S42, based on the leakage magnetic coupling coefficient matrix and winding geometric topology parameters, the Lorentz force under spatial leakage magnetic coupling is calculated on the gated effective short-circuit current vector to obtain the Lorentz force peak value prediction matrix. It should be noted that since the gated effective short-circuit current vector only represents the current quantity, while the elastic damping element needs to withstand the electrodynamic force generated by the current in the leakage magnetic field, the distribution of this force in different parts of the device body is not uniform, and the radial and axial action mechanisms are different. Based on this, the technical solution of this application further calculates the Lorentz force under spatial leakage magnetic coupling on the gated effective short-circuit current vector based on the leakage magnetic coupling coefficient matrix and winding geometric topology parameters to obtain the Lorentz force peak value prediction matrix. Through the above processing, the force amplitude distribution of each segment of the winding can be effectively given according to direction before mechanical vibration occurs, providing a quantitative basis for the target quantity and direction of the excitation current.

[0042] More specifically, in a specific example of this application, the winding of each phase is first discretized into several blocks according to the axial height and the number of radial layers. Each block is described by its number of turns, average diameter, radial thickness, and axial height. These parameters constitute the winding geometric topology parameters and are pre-stored in memory. Then, the peak current of the corresponding phase in the gated effective short-circuit current vector is distributed according to the number of turns of each block to obtain the current element of each block. The leakage magnetic coupling coefficient matrix is ​​multiplied by the current vector to deduce the leakage magnetic density vector at the location of each block. This coefficient matrix is ​​calibrated by the self-inductance and mutual inductance parameters of the winding through finite element field-circuit coupling calculation, and its output is divided into axial and radial components. Then, the current element of each block is cross-multiplied by the leakage magnetic density vector at its location to obtain the electrodynamic vector of that block, the expression of which is:

[0043] in, Indicates the first The electrodynamic vector of each segmented winding, Indicates the block index number. This indicates the number of turns in the block. This represents the equivalent conductor length of the segment along the circumference, and its value is the product of the average diameter of the segment and pi. This represents the block current vector obtained by distributing the gated effective short-circuit current vector according to the number of turns, with its direction along the circumference of the winding. This represents the leakage magnetic flux density vector at the location of the block, derived from the leakage magnetic coupling coefficient matrix. It contains both axial and radial components. This represents the vector cross product operation. The cross product result in the circumferential current interacting with the axial leakage magnetic component to produce a radial force, and the circumferential current interacting with the radial leakage magnetic component to produce an axial force. The directions of the two are determined by the right-hand rule, so this formula retains the direction of the force while giving the force amplitude.

[0044] Based on this, the electrodynamic vectors of each block are decomposed radially and axially and arranged according to the block index to form the peak Lorentz force prediction matrix, the expression of which is:

[0045] in, This represents the calculated peak Lorentz force prediction matrix, where the number of rows equals the total number of blocks and the number of columns equals the number of directional force components. Indicates the first The radial component of the electrodynamic vector acting on each block This represents the axial component of the vector. This matrix represents the total number of blocks. It is organized with rows corresponding to spatial locations and columns corresponding to the direction of action, recording both the force amplitude and direction of each block. The radial component characterizes the degree of compression in the inner winding and tension in the outer winding, while the axial component characterizes the compressive or dispersive tendency between the coils. Since the gated effective short-circuit current vector is zero under steady-state conditions, all elements of this matrix are zero during the transformer's steady-state operation at power frequency, and the feedforward path does not output force prediction results. When an external short circuit occurs on the primary side, this matrix, along with the spatial coordinates of each block, is output for subsequent stress wave propagation delay deduction and reverse current calculation, respectively.

[0046] In step S43, based on the spatial coordinates of the force application point in the Lorentz force peak prediction matrix and the preset wave velocity constant of the winding material and the sound velocity constant of the insulating oil, the stress wave cross-medium propagation path is time-delayed to obtain the stress wave arrival time difference. It should be noted that since the Lorentz force peak prediction matrix only provides the force amplitude distribution and not the arrival time of the stress wave excited by this force at the elastic damping element, the available time window for excitation and magnetization cannot be determined. Therefore, the technical solution of this application further uses the spatial coordinates of the force application point in the Lorentz force peak prediction matrix and the preset wave velocity constant of the winding material and the sound velocity constant of the insulating oil to perform time-delayed calculations of the stress wave cross-medium propagation path to obtain the stress wave arrival time difference. Through the above processing, the available time window for feedforward control can be effectively quantified, matching the application rate of the excitation current rising edge with this time window.

[0047] More specifically, in a particular example of this application, from the Lorentz force peak prediction matrix, each block is sorted according to the combined amplitude of the radial and axial components of each block. The block with the largest combined amplitude and its spatial coordinates are extracted. The shortest propagation path of the stress wave is constructed using these coordinates and the coordinates of the installation position of the elastic damping element. The combined amplitude is obtained by taking the square root of the sum of the squares of the radial and axial components of the block. The path is then divided into two segments according to the dielectric interface. One segment is the equivalent path length of the stress wave propagating through the winding conductor, pad, and solid insulation. The other segment is the stress wave... The equivalent path length traversed by the transformer insulating oil to reach the damping element at the bottom of the tank is calculated. The lengths of these two segments are pre-defined in the memory constant area based on the geometric dimensions in the transformer assembly drawings. Then, the pre-set wave velocity constant of the winding material and the sound velocity constant of the insulating oil are read. The former corresponds to the propagation speed of mechanical stress waves in the solid material of copper or aluminum windings, and the latter corresponds to the propagation speed of sound waves in the liquid insulating oil medium. The segmented propagation time is obtained by quotienting the path length of each segment with the wave velocity of the medium in which it is located, and then summed. The summation result is then gated using the impact condition trigger flag as a multiplication coefficient. The expression is as follows:

[0048] in, This represents the time difference of arrival of the stress wave obtained from the deduction. This indicates the triggering identifier for the impact condition, which serves as the gating multiplication coefficient. This represents the equivalent path length of stress wave propagation within the windings and solid insulation components. This represents the preset wave velocity constant of the winding material. This represents the equivalent path length of the stress wave propagating through the insulating oil to the damping element. This represents the preset sound velocity constant of the insulating oil. The two quotients in the formula are the propagation time of the stress wave in the solid medium section and the liquid medium section, respectively. Their sum is the total propagation time delay from the point of application of the force to the elastic damping element. This time delay is on the order of milliseconds, while the feature extraction, limit judgment, and mechanical calculation on the electrical side are completed on the order of microseconds within the gate array. The difference between the two constitutes the time window available for magnetization of the excitation coil. When the impact condition trigger flag is cleared, this time difference is forcibly clamped to zero, and the feedforward path does not output timing constraints. When the flag is set, the stress wave arrival time difference and the Lorentz force peak value prediction matrix are output together. Subsequently, the rising edge gradient constraint of magnetization is applied to the target current obtained by inverse solution based on this.

[0049] Specifically, in step S5, a dual-mode switching is performed based on the impact condition trigger flag to obtain the solution result of the current effective mode, and the solution result of the current effective mode is synthesized into the excitation current target matrix. It should be noted that, given that the two types of mechanical excitations borne by the transformer are different in cause and frequency band, the steady-state micro-vibration caused by the magnetostriction of the iron core needs to be continuously suppressed in a closed loop based on the measured mechanical state, while the transient impact caused by the external short circuit needs to be applied in advance based on the prediction result of the electrical side. If a single control path is used to deal with both, the feedforward solution result and the feedback solution result cannot coexist in the excitation coil at the same time, and the excitation winding of the magnetorheological elastic damping element only accepts a single current command. Based on this, the technical solution of this application further performs dual-mode switching based on the impact condition trigger flag to obtain the calculation result of the current effective mode, and synthesizes the calculation result of the current effective mode into an excitation current target matrix. That is, the flag and its complement value are used as complementary weights of the feedforward path and the feedback path, respectively. When the flag is set, the feedforward pre-bias target current component calculated by the Lorentz force peak value prediction matrix and the stress wave arrival time difference is taken. When the flag is cleared, the feedback steady-state adjustment target current component calculated by the net mechanical vibration state matrix is ​​taken. After weighted summation, a unique current command is formed. Through the above processing, the two control targets of impact resistance and steady-state vibration reduction can be effectively implemented on the same execution channel according to the working condition. During the impact, stiffness hardening is used to resist the electrodynamic peak, and after the impact subsides, the suppression of magnetostrictive micro-vibrations is restored, thereby taking into account the wide-band vibration reduction performance while avoiding command conflicts.

[0050] Figure 5 This is a flowchart illustrating a method for shock resistance and vibration reduction of a transformer with elastic damping elements according to an embodiment of this application. It describes a method for dual-mode switching based on an impact condition trigger flag to obtain the calculation result of the current effective mode, and then synthesizing the calculation result of the current effective mode into an excitation current target matrix. (See flowchart for example.) Figure 5As shown, step S5 includes: S51, under the nanosecond-level pulse drive of the timing synchronization control clock signal, performing nonlinear yield stress reverse current solution on the Lorentz force peak prediction matrix, and applying magnetic rise edge gradient constraint using the stress wave arrival time difference to obtain the feedforward pre-bias target current component; S52, performing steady-state micro-vibration feedback current solution based on mechanical state on the net mechanical vibration state matrix to obtain the feedback steady-state adjustment target current component; S53, based on the impact condition triggering flag, performing dual-mode command smooth weighted fusion of the feedforward pre-bias target current component and the feedback steady-state adjustment target current component to obtain the excitation current target matrix.

[0051] In step S51, under the nanosecond-level pulse drive of the timing synchronization control clock signal, the peak Lorentz force prediction matrix is ​​solved nonlinearly by yield stress inverse current, and the rising edge gradient constraint of the magnetization is applied using the stress wave arrival time difference to obtain the feedforward pre-bias target current component. It should be noted that, because the yield stress of the magnetorheological damping element has a nonlinear relationship with the excitation current, the peak Lorentz force prediction matrix cannot be directly converted into a current command in a linear proportion, and the magnetization process of the excitation coil must be completed before the stress wave arrives. Based on this, the technical solution of this application further performs a nonlinear yield stress inverse current solution on the peak Lorentz force prediction matrix under the nanosecond-level pulse drive of the timing synchronization control clock signal, and applies the rising edge gradient constraint of the magnetization using the stress wave arrival time difference to obtain the feedforward pre-bias target current component. Through the above processing, the target amount of excitation current required for shock resistance and its allowable rise rate can be effectively determined within the time window.

[0052] More specifically, in a specific example of this application, a nanosecond-level pulse of the timing synchronization control clock signal is used as the hard real-time activation condition for the operation. When the effective edge of the pulse is detected, the inverse solution is initiated. The radial and axial components of each segment winding in the Lorentz force peak prediction matrix are substituted into the nonlinear yield stress inverse function of the magnetorheological material. This inverse function is calibrated according to the Bingham plastic constitutive relation and describes the characteristic that the yield stress tends to saturate as the excitation current increases in exponential form. The radial component corresponds to the hardening stiffness required for the horizontal relative displacement between the constraint body and the oil tank, and the axial component corresponds to the hardening stiffness required for the vertical relative displacement. Thus, the theoretical target current matrix that enables the elastic damping element to reach the hardening stiffness required to resist the peak value of the component is calculated in reverse according to the direction. The radial and axial current components are then distributed to the corresponding excitation coils according to the installation direction of the elastic damping element. This inverse solution is implemented in the gate array by a lookup table combined with pipelined multiplication and addition, and the time consumption is on the order of microseconds, which is negligible compared to the propagation delay of the stress wave.

[0053] Based on this, the arrival time difference of the stress wave is read. The remaining time after deducting the computational time of each step (feature extraction, limit judgment, mechanical prediction, and reverse solution) is used as the available time window for magnetization. A rising edge gradient constraint function is then constructed to ensure that the rise slope of the target current satisfies the magnetization rate constraint determined by the inductance of the excitation coil and the output capability of the constant current driver. Its expression is:

[0054] in, This represents the feedforward pre-bias target current component obtained from the solution. This indicates the impact condition trigger flag, which serves as the gating factor. Its value is either 1 or 0, and it is used for timing synchronization control clock signals. It does not appear in this formula; its function is to trigger the inverse calculation operation on the first valid pulse edge after the flag is set. The calculation result is latched and held in the register until the next control cycle when it is recalculated based on the new input. This keeps the target current component at a stable value and prevents ripple from occurring with the on and off of the clock pulse. The inverse material constitutive constant matrix represents the magnetorheological elastic damping element. This represents the Hadamard product, which is the product of corresponding elements of a matrix. This represents the peak Lorentz force prediction matrix. The inverse exponential constant representing the nonlinear yielding effect. This represents the rising edge safety time window constraint function. This represents the time difference of arrival of the stress wave. In the formula... The theoretical target current distribution, calculated inversely based on the nonlinear constitutive relation, compensates for the saturation characteristics of yield stress variation with current in a power-law form. This distribution is then multiplied by a constraint factor determined by the arrival time difference of the stress wave, so that the actual current command issued meets the impact resistance requirements in terms of amplitude and falls within the available time window in terms of timing.

[0055] To ensure that the magnetization process can be completed within the aforementioned time window, the constant current driver does not output at the rated voltage during the initial stage of magnetization. Instead, it applies a strong excitation voltage higher than the rated value, which compresses the current rise time constant of the excitation coil from its natural time constant. The time required for the excitation current to rise from zero to the target value is on the order of milliseconds, which is less than the stress wave propagation delay from the point of application of the winding force to the elastic damping element. When the measured current approaches the target value, the driver exits the strong excitation and enters the constant current maintenance state to avoid local magnetic saturation of the magnetorheological elastomer caused by current overshoot. When the transformer winding structure and damping element installation position result in a small time difference between the arrival of the stress wave and the actual arrival time, such that the remaining time after deducting the computation time is still less than the magnetization time required under strong excitation conditions, the rising edge gradient limiting function degenerates into amplitude truncation. This means that a command is issued based on the upper limit of the excitation current achievable within the available time window, ensuring that the element possesses the hardening stiffness corresponding to that current value at the arrival time of the first wave. After the first wave arrives, the current is continuously supplemented to the theoretical target value according to the limited slope to withstand the electrodynamic force generated by subsequent cycles of the short-circuit current. After the solution is completed, the feedforward pre-biased target current component is output to the dual-mode fusion stage, and its weight is assigned to the solution results from the feedback path according to the impact condition trigger flag.

[0056] In step S52, the net mechanical vibration state matrix is ​​calculated based on the steady-state micro-vibration feedback current to obtain the feedback steady-state adjustment target current component. More specifically, in a specific example of this application, the displacement deviation vector and velocity deviation vector at the installation position of the elastic damping element are extracted from the net mechanical vibration state matrix. These two vectors are obtained by subtracting the posterior state estimation result from the target vibration state, where the target vibration state is set according to the transformer noise and vibration factory limits. Subsequently, the above deviation vectors are sent to a digital proportional-integral-derivative closed-loop controller. The proportional path generates an instantaneous response based on the current deviation amplitude, the integral path weights the cumulative value of the deviation over several control cycles to eliminate the periodic steady-state error caused by magnetostriction, and the derivative path provides advance suppression based on the changing trend of the deviation. The sum of these three is mapped to a damping force command for counteracting the power frequency harmonic micro-vibration, the expression of which is:

[0057] in, This represents the feedback steady-state damping force command obtained from the solution. , and Let these represent the proportional gain matrix, integral gain matrix, and differential gain matrix, respectively. This represents the mechanical error state vector for the current control cycle extracted from the net mechanical vibration state matrix. Indicates the control cycle index within the integral interval. Indicates the current control cycle number. Indicates the first Mechanical error state vector for each control cycle This indicates the sampling time interval between adjacent control cycles. This represents the difference between the mechanical error state vector of the current control cycle and the previous control cycle. The second term in the formula is the discrete integral accumulation of the error, used to compensate for the periodic deviation of the magnetostrictive vibration; the third term is the discrete derivative of the error, used to suppress rapid changes in the vibration state. The obtained damping force command is then converted from the force domain to the current domain via the electromagnetic force positive calibration curve of the magnetorheological elastic damping element, obtaining the feedback steady-state adjustment target current component, and output to the dual-mode fusion stage according to the weight allocation of the compensation value of the impact condition trigger identifier.

[0058] The above example uses a single set of proportional-integral-derivative (PID) parameters to uniformly process all vibration components in the net mechanical vibration state matrix, suitable for scenarios where the transformer operates in a long-term steady state and the matrix does not contain residual energy after an impact. It should be noted that the above example directly extracts displacement and velocity deviations from the net mechanical vibration state matrix and uniformly feeds all deviations into the same set of PID parameters, which is predicated on the vibration states in the matrix being stable and homogeneous. However, during the transition from short-circuit impact condition to steady-state operation, the structural modal oscillations in the windings, clamps, tank, and elastic damping elements that have not yet dissipated continue. This causes the net mechanical vibration state matrix to simultaneously contain steady-state harmonic components formed by core magnetostriction and transient residual components formed by impact attenuation. The frequency ranges, attenuation patterns, and control requirements of these two components are different, and the impact condition triggering flag can only indicate whether the impact judgment condition is currently met, not whether the residual vibration has completely attenuated. Therefore, this application further provides the following enhanced embodiment.

[0059] Figure 6 This document presents a flowchart illustrating the modal separation of the net mechanical vibration state matrix and the calculation based on the differences between the steady-state harmonic channel and the impact residual channel to obtain the feedback steady-state adjustment target current component, according to an embodiment of this application for a transformer shock resistance and vibration reduction method with elastic damping elements. Figure 6 As shown, step S52 includes: S521, based on the impact condition triggering identifier, performing vibration matrix adaptive mode separation on the net mechanical vibration state matrix to obtain the magnetostrictive steady-state harmonic component and the impact attenuation transient residual component; S522, performing dual-channel mode discrimination weighted control force calculation on the magnetostrictive steady-state harmonic component and the impact attenuation transient residual component to obtain the mode discrimination weighted damping force command; S523, through a preset magnetorheological material electromagnetic force positive calibration lookup table, performing piecewise linear interpolation mapping transformation from the force domain to the current domain on the damping force value of each channel in the mode discrimination weighted damping force command to obtain the feedback steady-state adjustment target current component.

[0060] In step S521, based on the impact condition triggering flag, adaptive mode separation of the vibration matrix is ​​performed on the net mechanical vibration state matrix to obtain the magnetostrictive steady-state harmonic components and the impact attenuation transient residual components. It should be noted that, since the impact condition triggering flag can only indicate whether the impact judgment condition is currently met, and cannot indicate whether the mechanical energy in the transformer windings, clamps, tank, and elastic damping elements has been completely dissipated, immediately processing the net mechanical vibration state matrix in steady-state mode after the flag changes from 1 to 0 will mistakenly treat the low-frequency amplitude residual vibration that has not yet attenuated as steady-state control error. This residual vibration is not sensor noise, and Kalman filtering cannot filter it out without losing the true mechanical state. Therefore, the technical solution of this application further performs adaptive mode separation of the vibration matrix on the net mechanical vibration state matrix based on the impact condition triggering flag to obtain the magnetostrictive steady-state harmonic components and the impact attenuation transient residual components. Through the above processing, the true steady-state operating range and the modal transition range after the impact can be effectively distinguished, establishing clear data boundaries for vibration components formed by different physical mechanisms.

[0061] More specifically, in a specific example of this application, in addition to the net mechanical vibration state matrix and the impact condition trigger flag of the current control cycle, a historical state sequence of the flag is input separately. Falling edge detection is performed on the historical state sequence to determine the impact exit moment when the flag jumps from 1 to 0, and the control cycle is accumulated from that moment. Based on this, the accumulated number of control cycles is compared with a preset transient decay cycle threshold. This threshold is calibrated according to the decay time constant of the transformer body structural mode and is used to define the time span of the mode transition interval. When the accumulated number of control cycles is less than the preset transient decay cycle threshold, an adaptive cutoff frequency is determined according to the transformer structural mode calibration parameters and the accumulated number of control cycles. The complementary digital filter corresponding to the cutoff frequency is used to perform frequency domain mode separation on the net mechanical vibration state matrix to obtain the magnetostrictive steady-state harmonic component and the impact decay transient residual component. When the accumulated number of control cycles reaches the transient decay cycle threshold, the net mechanical vibration state matrix is ​​used as the magnetostrictive steady-state harmonic component, and the impact decay transient residual component is set to zero, thereby avoiding unnecessary dual-channel calculations during the normal steady state period. The adaptive cutoff frequency adjusts as the cumulative number of cycles increases, ensuring the boundary position aligns with the actual decay process of the residual vibration gradually converging to lower frequencies during the transition period. The separation relationship during the modal transition is expressed as:

[0062]

[0063] in, This represents the magnetostrictive steady-state harmonic component, which is the vibrational component caused by the magnetostrictive effect of the iron-core silicon steel sheet and concentrated in the power frequency harmonics and their higher harmonics. This represents the transient residual component of the impact attenuation, that is, the low-frequency vibration component remaining after the modal oscillation of the vessel structure has not yet dissipated after the impact has subsided. Represents the net mechanical vibration state matrix. This indicates that digital filter operators with frequencies higher than the adaptive cutoff frequency are retained. This indicates a digital filter operator that is complementary to the aforementioned digital filter operators and retains a frequency not higher than the adaptive cutoff frequency. Indicates the vibration frequency. This represents the adaptive cutoff frequency corresponding to the nth cumulative control cycle, where n represents the number of cumulative control cycles since the impact exit moment. It should be noted that the passband and stopband of the two types of filter operators are complementary, ensuring that the two separated components can jointly characterize the effective mechanical state in the net mechanical vibration state matrix, thus avoiding the unnecessary loss of vibration state energy due to frequency domain separation.

[0064] In a more specific application scenario, after a large power transformer experiences an external short circuit, the impact condition trigger flag remains at 1 for several control cycles before jumping to 0. At this point, the falling edge detection determines the impact exit moment and begins cycle accumulation. Since the structural modal oscillations of the clamping components and the elastic damping element at the bottom of the tank are still ongoing, the accumulated cycle number has not yet reached the transient decay cycle threshold. Based on the structural modal calibration parameters and the adaptive cutoff frequency determined by the current accumulated cycle number, the net mechanical vibration state matrix is ​​divided into two parts. The power frequency harmonics and their higher harmonics above this frequency are assigned to the magnetostrictive steady-state harmonic components, while the low-frequency amplitude oscillation components not higher than this frequency are assigned to the impact decay transient residual components. As the residual energy gradually dissipates, after the accumulated cycle number reaches the threshold, the impact decay transient residual components are set to zero. The net mechanical vibration state matrix as a whole participates in subsequent calculations as the magnetostrictive steady-state harmonic components, and the modal separation operation is then terminated.

[0065] In step S522, a dual-channel modal discrimination weighted control force calculation is performed on the magnetostrictive steady-state harmonic component and the impact-damped transient residual component to obtain a modal discrimination weighted damping force command. It should be noted that when the magnetostrictive steady-state harmonic component and the impact-damped transient residual component are uniformly fed into the same set of proportional-integral-derivative (PID) parameters, the integral path will continuously accumulate the large-amplitude deviation caused by residual vibration, resulting in integral saturation and recovery hysteresis. The derivative path will identify the rapid decay edge of residual vibration as a sudden change in mechanical state, generating abnormal current spikes. Simply using integral limiting or switching control parameters still adjusts the gain on the aliased signal. Based on this, the technical solution of this application further performs a dual-channel modal discrimination weighted control force calculation on the magnetostrictive steady-state harmonic component and the impact-damped transient residual component to obtain a modal discrimination weighted damping force command. Through the above processing, the configuration of the control path can be effectively matched with the physical characteristics of the two types of vibration, suppressing erroneous integral and abnormal derivative responses after impact exit while maintaining the accuracy of steady-state closed-loop control.

[0066] More specifically, in a specific example of this application, a harmonic error vector corresponding to the target vibration state is extracted from the magnetostrictive steady-state harmonic component. This target vibration state is set according to the transformer's factory vibration limit. Proportional, integral, and differential three-channel control operations are performed on the harmonic error vector. The proportional path responds to the current harmonic deviation, the integral path eliminates the persistent periodic steady-state error, and the differential path suppresses the changing trend of the magnetostrictive vibration. The sum of these three constitutes the steady-state harmonic suppression force component. Simultaneously, only pure proportional damping operations are performed on the impact attenuation transient residual component, actively cutting off its integral and differential paths to avoid integral accumulation caused by low-frequency large-amplitude residual vibration and differential spikes caused by attenuation edges. The resulting transient residual attenuation force component applies additional energy dissipation along the attenuation direction of the residual vibration, and its amplitude decreases synchronously with the decrease of the residual vibration. The steady-state harmonic suppression force component generated by the harmonic channel and the transient residual attenuation force component generated by the residual channel are vector-synthesized to obtain the modal discrimination weighted damping force command, expressed as:

[0067] in, This indicates the modal discrimination weighted damping force command. , and Let these represent the proportional gain matrix, integral gain matrix, and differential gain matrix of the magnetostrictive steady-state harmonic channel, respectively. This represents the harmonic error vector for the k-th control cycle. Let represent the harmonic error vector of the j-th control cycle within the integration interval, where j represents the integral summation index and k represents the current control cycle number. This indicates the sampling time interval between adjacent control cycles. This represents the harmonic error vector of the previous control cycle. This represents the proportional damping gain matrix of the impact residual channel. This represents the transient residual component of the impact attenuation during the k-th control cycle. The first three terms constitute the complete proportional-integral-differential output of the harmonic channel. The integral term accumulates the harmonic error by weighting the sampling time interval to eliminate periodic deviations, and the differential term approximates the error change rate by dividing the difference of the harmonic errors of adjacent cycles by the sampling time interval. The last term is the pure proportional output of the residual channel. This term does not contain accumulation or differential operations, so the residual vibration neither enters the integral accumulation nor participates in the edge differential.

[0068] In a more specific application scenario, during the modal transition range after impact exit of a large power transformer, low-frequency large-amplitude oscillations still exist at the elastic damping element at the bottom of the tank. After modal separation and inclusion into the transient residual component of impact attenuation, this oscillation is only affected by the proportional damping gain matrix. As its amplitude naturally decays, it outputs a decreasing transient residual attenuation force component, and the excitation current does not continuously increase due to the accumulation of this oscillation in the integrator. At the same time, the power frequency harmonics and their higher harmonic components are included into the magnetostrictive steady-state harmonic components and maintain complete three-channel control. The steady-state micro-vibration suppression accuracy is not affected by the residual vibration during the transition period. This control allocation matches the physical characteristics of the two types of vibrations. Periodic magnetostrictive vibration requires the use of integral and derivative actions to improve tracking accuracy and dynamic suppression capability, while impact residual vibration only needs to provide additional energy dissipation along its attenuation direction, avoiding the controller from continuously accumulating control quantity due to the unreleased residual energy. Thus, it retains the steady-state closed-loop control accuracy while suppressing erroneous integral and abnormal derivative responses after impact exit.

[0069] In step S523, a piecewise linear interpolation mapping transformation from the force domain to the current domain is performed on the damping force values ​​of each channel in the modal discrimination weighted damping force command using a preset magnetorheological material electromagnetic force forward calibration lookup table to obtain the feedback steady-state adjustment target current component. It should be noted that, given the nonlinear relationship between the yield stress, equivalent stiffness, and damping coefficient of the magnetorheological material and the excitation current, directly converting the modal discrimination weighted damping force command into current using a fixed proportional coefficient would distort the aforementioned modal separation and differentiated control results in the force-to-electricity conversion stage. Therefore, the technical solution of this application further uses a preset magnetorheological material electromagnetic force forward calibration lookup table to perform a piecewise linear interpolation mapping transformation from the force domain to the current domain on the damping force values ​​of each channel in the modal discrimination weighted damping force command to obtain the feedback steady-state adjustment target current component. Through the above processing, the current command can be effectively matched with the required damping force, and the differentiated control results can be completely transmitted to the execution stage.

[0070] More specifically, in a specific example of this application, the input modal discrimination weighted damping force command is matched with the electromagnetic force calibration data of the magnetorheological elastic damping element under different temperatures, pre-compression amounts, and excitation current conditions. This calibration data is measured point by point during the component's factory test and stored in memory in the form of a multidimensional lookup table. The temperature dimension corresponds to the material property drift caused by transformer oil temperature and coil heating, and the pre-compression amount dimension corresponds to the static compression state of the component after bearing the weight of the transformer body. Then, the segment where the lookup table is located is located according to the current measured temperature and pre-compression amount. Within this segment, the two adjacent calibration nodes where it falls are determined by the damping force value as an index. Piecewise linear interpolation is performed between adjacent calibration nodes to convert the force domain control quantity into the current domain control quantity, generate the feedback steady-state adjustment target current component, and convert the damping force value corresponding to each channel in the command after the synthesis of the steady-state harmonic channel and the impact residual channel separately, so that the energy consumption required by the two types of vibrations maintains their respective proportional relationship in the current domain. By using a calibration lookup table instead of direct linear proportional conversion, the nonlinear relationship between the yield stress, equivalent stiffness, and damping coefficient of magnetorheological materials and the excitation current can be adapted, enabling the target current component of the feedback steady-state adjustment to accurately carry the aforementioned modal separation and differentiated control results.

[0071] Regarding the connection of data streams, after generating the feedback steady-state adjustment target current component, the feedforward pre-bias target current component and the impact condition triggering flag retain their original data names and are transmitted to the dual-mode command fusion stage through the bypass register channel. The fusion weight of the feedforward path and the feedback path is determined by the impact condition triggering flag.

[0072] In a more specific application scenario, during the modal transition range after the impact subsidence of a large power transformer, the oil temperature is in a relatively high range while the pre-compression of the elastic damping element remains at the rated assembly value. Based on this, the force-electric slope between adjacent nodes in the calibration segment is smaller than that of the normal temperature segment, and the interpolated current value is correspondingly increased to compensate for the decrease in yield stress caused by the temperature rise. As the residual vibration decays, the damping force contributed by the residual impact channel decreases, and the current component of the interpolated output drops synchronously. The excitation current does not exhibit repeated adjustments unrelated to the residual vibration, and the peak current of the constant current driver and the additional heating of the excitation coil are suppressed accordingly. The transition time of the transformer from the impact condition to the steady-state vibration reduction condition is shortened.

[0073] In step S53, based on the impact condition triggering identifier, the feedforward pre-bias target current component and the feedback steady-state adjustment target current component are subjected to dual-mode command smoothing weighted fusion to obtain the excitation current target matrix. It should be noted that, since the feedforward pre-bias target current component and the feedback steady-state adjustment target current component correspond to the impact condition and steady-state condition respectively, and the excitation coil of the magnetorheological elastic damping element only receives a single current command at any given time, the technical solution of this application further performs dual-mode command smoothing weighted fusion on the feedforward pre-bias target current component and the feedback steady-state adjustment target current component based on the impact condition triggering identifier to obtain the excitation current target matrix. Through the above processing, the two calculation results can be effectively made to take effect in a time-sharing manner according to the current operating condition, avoiding command conflicts.

[0074] More specifically, in a specific example of this application, the system receives the feedforward pre-bias target current component calculated by the feedforward path, the feedback steady-state adjustment target current component obtained by calibration and table lookup transformation from the feedback path, and an impact condition trigger flag with a value of 1 or 0. Using the flag itself as the weight of the feedforward path and the difference between 1 and the flag as the weight of the feedback path, a complementary mask is constructed where the sum of the two weights is always 1. This ensures that when an external short circuit occurs on the primary side, the weight of the feedforward path is 1 and the weight of the feedback path is 0; during the transformer's steady-state operation at power frequency, the two values ​​are reversed. Subsequently, the two current components are multiplied by their corresponding weights and matrix addition is performed to synthesize a unique excitation current target matrix, the expression of which is:

[0075] in, This represents the synthesized target matrix of excitation current. This indicates the triggering identifier for the impact condition, which serves as a complementary weighting coefficient. This represents the target current component for the feedforward pre-bias. This represents the target current component for feedback steady-state adjustment. The two terms in the formula are the weighted outputs of the feedforward and feedback paths, respectively. Due to the complementary weights, only one component contributes to the synthesis result at any given time, while the other is zeroed out. This allows mode selection to be achieved by replacing conditional jumps with algebraic operations, avoiding scheduling delays introduced within millisecond time windows due to branch judgments. Furthermore, to prevent current surges in the excitation coil caused by step jumps in the current command during mode switching, a slope limit is applied to the synthesis result within several control cycles during which the weights flip. This ensures the command transitions to the target value of the new mode along an allowable ramp or descent slope. The upper limit of the slope in the impact condition intervention direction is constrained by the stress wave arrival time difference, while the upper limit of the slope in the steady-state recovery direction is set according to the inductance of the excitation coil and the output capability of the constant current driver. The synthesized excitation current target matrix is ​​output to the power amplification stage to drive the magnetorheological damping element to complete stiffness and damping reconstruction.

[0076] Specifically, in step S6, the excitation current target matrix is ​​amplified and injected into the excitation coil of the magnetorheological elastic damping element for stiffness and damping reconstruction. It should be noted that since the excitation current target matrix is ​​a weak electrical command output by the controller, its power is insufficient to establish the required magnetic field in the excitation coil of the magnetorheological elastic damping element, and the calculation results of the aforementioned steps cannot be directly applied to the transformer body. Based on this, the technical solution of this application further amplifies the excitation current target matrix and injects it into the excitation coil of the magnetorheological elastic damping element for stiffness and damping reconstruction. Through the above processing, the current command can be effectively converted into actual changes in the mechanical characteristics of the element, allowing the control results to be implemented in the vibration transmission path of the transformer.

[0077] More specifically, in a particular example of this application, the synthesized excitation current target matrix is ​​divided into several independent commands according to the installation number of each elastic damping element, and sent to a constant current driver located at the bottom of the transformer. This driver adopts a bidirectional DC-DC converter topology and operates in pulse width modulation mode. Its inner current loop is given by the command value and fed back by the measured value of the Hall current sensor. Based on this, the duty cycle of the switching device is adjusted so that the actual current in the excitation coil tracks the command value and does not drift with the change of DC resistance and back electromotive force caused by the temperature rise of the coil. The amplified current is injected into the excitation coil arranged on both sides of the magnetorheological elastomer, establishing a magnetic field in the working gap of the element, causing the magnetic particles in the magnetorheological elastomer to form chains along the direction of the magnetic field. The equivalent shear modulus and damping coefficient of the element change accordingly, thereby completing the stiffness and damping reconstruction. When an external short circuit occurs on the primary side, the injected current, with an amplitude determined by the feedforward pre-bias target current component, completes magnetization before the stress wave arrives. The component stiffness rises to the level required to resist the peak value of the Lorentz force, and the relative displacement between the transformer body and the tank is limited, preventing the elastic element from being compressed to its physical limit and triggering a secondary rigid collision. During the transformer's steady-state operation at power frequency, the injected current is continuously fine-tuned according to the feedback steady-state adjustment target current component, and the energy dissipation characteristics of the component are adapted to the micro-amplitude vibration caused by the magnetostriction of the iron core. In addition, the constant current method rather than the constant voltage method is used for driving, so that the magnetic field strength on the magnetorheological elastic damping element maintains a corresponding relationship with the solution result. The conversion accuracy of the aforementioned nonlinear yield stress inverse solution and electromagnetic force calibration lookup table is retained in the execution stage.

[0078] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for shock resistance and vibration reduction of a transformer with an elastic damping element, characterized in that, include: S1, acquire the three-phase transient current signal on the primary side of the transformer, the mechanical acceleration signal of the transformer body tank, and historical mode operating parameters; S2, perform wavelet transient feature extraction on the three-phase transient current signal of the primary side of the transformer to obtain the transient electrical feature vector, and perform Kalman filtering on the tank mechanical acceleration signal based on the historical mode operating parameters to obtain the net mechanical vibration state matrix; S3, perform energy limit determination on transient electrical feature vectors to obtain impact condition triggering identifier; S4, in response to the effective setting of the impact condition triggering flag, performs cross-domain feedforward mechanical prediction calculation on the transient electrical characteristic vector to obtain the Lorentz force peak prediction matrix and stress wave arrival time difference; S5, based on the impact condition triggering flag, performs dual-mode switching to obtain the solution result of the current effective mode, and synthesizes the solution result of the current effective mode into the excitation current target matrix; S6, after power amplification, injects the target matrix of excitation current into the excitation coil of the magnetorheological elastic damping element to reconstruct stiffness and damping.

2. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 1, characterized in that, Step S2 includes: Based on a global hardware clock, the three-phase transient current signal and the tank mechanical acceleration signal are synchronously converted and acquired to obtain a discretized three-phase current sequence and a discretized acceleration sequence. The historical mode operating parameters are unpacked to extract the process noise covariance. Wavelet domain electrical transient feature decomposition is performed on the discretized three-phase current sequence to obtain transient electrical feature vectors; The Kalman filter is initialized with prior covariance weights based on the process noise covariance, and the discretized acceleration sequence is subjected to state recursion and measurement update filtering to obtain the net mechanical vibration state matrix.

3. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 2, characterized in that, Wavelet domain electrical transient feature decomposition is performed on the discretized three-phase current sequence to obtain transient electrical feature vectors, including: Discrete wavelet transform is performed on the discretized three-phase current sequence to obtain high-frequency detail coefficients; The sum of squares of the high-frequency detail coefficients is calculated within a sliding window to obtain the absolute value of the transient energy, and the slope abrupt rate of the discretized three-phase current sequence is calculated. The transient energy absolute value and the slope mutation rate are concatenated to obtain the transient electrical feature vector.

4. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 1, characterized in that, Step S3 includes: The transient electrical feature vectors are mapped to the electrical mutation energy space to obtain the absolute value of the electrical mutation energy; The absolute value of electrical sudden change energy is compared with the preset insulation safety tolerance threshold, and the binary judgment is performed through step logic to obtain the impact condition triggering identifier. In response to the high-level setting of the impact condition trigger flag, the nanosecond-level oscillator inside the gate array is gated and enabled to generate a timing synchronization control clock signal.

5. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 4, characterized in that, To obtain the absolute value of electrical mutation energy, a spatial mapping of the transient electrical eigenvectors is performed, including: Extract the transient energy absolute value component and slope change rate component from the transient electrical feature vector, and assign normalized weight coefficients to the transient energy absolute value component and slope change rate component; Based on the normalized weighting coefficients, the absolute value of transient energy and the slope mutation rate are mapped in the second norm space to obtain the absolute value of electrical mutation energy.

6. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 1, characterized in that, Step S4 includes: Using the impact condition triggering identifier as the gating multiplication coefficient, transient feature gating modulation and short-circuit current reconstruction are performed on the transient electrical feature vector to obtain the gating effective short-circuit current vector. Based on the leakage magnetic coupling coefficient matrix and winding geometric topology parameters, the Lorentz force under spatial leakage magnetic coupling is calculated on the gated effective short-circuit current vector to obtain the Lorentz force peak prediction matrix. Based on the spatial coordinates of the force application point in the Lorentz force peak prediction matrix and the preset winding material wave velocity constant and insulating oil sound velocity constant, the time delay of the stress wave propagation path across the medium is extrapolated to obtain the stress wave arrival time difference.

7. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 1, characterized in that, Step S5 includes: Driven by nanosecond-level pulses of the timing synchronization control clock signal, the peak value prediction matrix of Lorentz force is solved by nonlinear yield stress reverse current, and the magnetization rising edge gradient constraint is applied by using the stress wave arrival time difference to obtain the feedforward pre-bias target current component. The net mechanical vibration state matrix is ​​calculated based on the steady-state micro-vibration feedback current to obtain the feedback steady-state adjustment target current component; Based on the impact condition triggering identifier, the feedforward pre-bias target current component and the feedback steady-state adjustment target current component are smoothly weighted and fused in a dual-mode command to obtain the excitation current target matrix.

8. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 1, characterized in that, Step S6 includes: amplifying the excitation current target matrix with power through a constant current driver and injecting it into the excitation coil of the magnetorheological damping element.

9. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 8, characterized in that, The excitation coil, which amplifies the excitation current target matrix through a constant current driver and injects it into a magnetorheological damping element, includes: According to the installation number of each magnetorheological elastic damping element, the excitation current target matrix is ​​split into multiple current commands; Using the current commands of each circuit as given values ​​and the actual current of the corresponding excitation coil as feedback values, the constant current driver is controlled in a closed loop. The power-amplified current is then injected into the excitation coil of the corresponding magnetorheological elastic damping element through the constant current driver.

10. The method for shock resistance and vibration reduction of a transformer with elastic damping elements according to claim 7, characterized in that, The net mechanical vibration state matrix is ​​used to perform steady-state micro-vibration feedback current calculation based on the mechanical state to obtain the feedback steady-state adjustment target current component, including: Based on the impact condition triggering identifier, the net mechanical vibration state matrix is ​​subjected to vibration matrix adaptive mode separation to obtain the magnetostrictive steady-state harmonic components and the impact attenuation transient residual components. A dual-channel modal discrimination weighted control force is calculated for the magnetostrictive steady-state harmonic components and the impact attenuation transient residual components to obtain the modal discrimination weighted damping force command; By using a pre-set lookup table for the electromagnetic force of magnetorheological materials, the damping force values ​​of each channel in the modal discrimination weighted damping force command are transformed from the force domain to the current domain through piecewise linear interpolation mapping to obtain the feedback steady-state adjustment target current component.

Images