Transformer self-adaptive vibration reduction system based on magnetorheological damper and multi-mode cooperative control method

Abstract

The invention discloses a transformer self-adaptive vibration reduction system based on a magnetorheological damper and a multi-mode cooperative control method, which are used for solving the problems of insufficient low-frequency vibration suppression and poor environmental tolerance in the prior art. The vibration reduction system comprises a sensor module, a control unit, an execution mechanism and an energy recovery module. The sensor module monitors vibration signals of a transformer iron core and a winding in real time; the control unit converts the vibration signal into a magnetic field intensity adjusting instruction in real time, and the damping force accurately matches the target demand by actively compensating disturbance and dynamically optimizing parameters; the execution mechanism comprises a dual-channel magnetorheological damper, the energy recovery module comprises a friction nanometer generator TENG and a super capacitor, the output voltage of the TENG is stored in the super capacitor after DC-DC conversion, and energy closed-loop utilization is achieved. The problems that a traditional vibration reduction device is insufficient in low-frequency suppression, prone to corrosion and high in energy consumption are solved through multi-frequency-band vibration suppression, environment self-adaption design and the self-power-supply technology.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer vibration reduction, and in particular to a transformer adaptive vibration reduction system based on a magnetorheological damper and a multi-modal coordinated control method. Background Art

[0002] Traditional transformer vibration reduction devices have the following technical bottlenecks: (1) Insufficient low-frequency vibration suppression: The main frequency of transformer electromagnetic vibration is concentrated in 50Hz-200Hz. Traditional passive vibration isolators, such as rubber pads and spring isolators, cannot dynamically adapt to the vibration characteristics under different working conditions due to their fixed natural frequency. Studies have shown that the vibration isolation efficiency of rubber pads in the frequency band above 100Hz decreases by more than 40%. (2) Poor environmental adaptability: The high humidity and salt spray environment in scenarios such as offshore substations can easily cause corrosion of metal components and seal failure. For example, the stiffness of the spring isolator of a coastal substation decreased by 30% due to rust, causing resonance problems. (3) High energy consumption: The existing active vibration reduction system relies on external power supply, and the long-term operating cost is high. Taking a 500kV converter station as an example, the annual power consumption of the active vibration reduction system reaches 12,000 kWh, accounting for 15% of the total operation and maintenance cost.

[0003] To address the above problems, there is an urgent need for an intelligent vibration reduction system that integrates adaptive vibration reduction, environmental tolerance optimization and energy self-supply. Summary of the Invention

[0004] The purpose of the present invention is to provide a transformer adaptive vibration reduction system and a multi-modal collaborative control method based on magnetorheological dampers, which are used to solve the problems of insufficient low-frequency vibration suppression, poor environmental tolerance and high energy consumption in existing transformer vibration reduction systems.

[0005] The technical solution adopted by the present invention to solve its technical problems is: a transformer adaptive vibration reduction system based on a magnetorheological damper, including a sensor module, a control unit, an actuator and an energy recovery module; the sensor module is used to monitor the vibration acceleration and strain of the transformer core and winding in real time; the control unit converts the vibration acceleration signal into a magnetic field strength adjustment instruction in real time, and through active compensation of disturbances and dynamic optimization of parameters, the damping force is accurately matched to the target requirements; the actuator includes a dual-channel magnetorheological damper, the main channel performs low-frequency suppression, and the auxiliary channel performs high-frequency suppression; the energy recovery module includes a friction nanogenerator TENG and a supercapacitor, the friction nanogenerator TENG is embedded in the damper piston rod, and adopts a multi-layer triboelectric structure PTFE-Cu composite film, and the output voltage of the friction nanogenerator TENG is stored in the supercapacitor after DC-DC conversion, realizing closed-loop energy utilization.

[0006] Furthermore, the sensor module includes an acceleration sensor and a fiber grating strain sensor.

[0007] Furthermore, the control unit includes a tracking differentiator TD, an extended state observer ESO and a nonlinear feedback NLSEF.

[0008] Furthermore, the main channel adopts a spiral flow channel design with a spiral flow channel diameter of 8 mm, a Reynolds number Re<2000, and a pressure drop ΔP≤0.2 MPa.

[0009] Furthermore, the opening degree of the rotary valve of the auxiliary channel is 0-90 degrees, and the turbulence intensity of the liquid flow is adjusted by the rotary valve to suppress vibrations above 200 Hz.

[0010] Furthermore, the tracking differentiator TD generates a smooth target signal v1(t) and its differential v2(t). The algorithm formula of the tracking differentiator is: (1); where v1(k) and v2(k) are the state variables of the tracking differentiator; h is the control step size; Fhan is the fastest control synthesis function used to generate the control signal; F target is the target vibration signal; r and h0 are control parameters.

[0011] Furthermore, the extended state observer (ESO) expands the total system disturbance, including external vibration interference, the nonlinear characteristics of the magnetorheological fluid, and the time-varying characteristics of the damper parameters, into the third state variable z3(k) of the system. The algorithm formula of the extended state observer (ESO) is: (2); where z1(k), z2(k), and z3(k) are the state variables of the extended state observer; β 01 , β 02 , β 03 is the observer gain parameter; fal is the nonlinear error function used to suppress high-frequency noise; e1(k)=Z1-F actual , is the observed value Z1 of the damping force and the actual value F of the damping force actual b0 is the control gain, which characterizes the influence of unit current on the increment of damping force or equivalent magnetic field intensity; a1 is the power exponential parameter of the nonlinear error function fal, which determines the intensity of the nonlinear characteristic; δ is the linear interval width parameter of the nonlinear error function fal, which is used to control the segmented range of the function; u(k) is the control input signal, that is, the magnetic field coil current command value output at time k.

[0012] The present invention also provides a transformer adaptive vibration reduction multi-mode collaborative control method based on magnetorheological dampers, which includes the following steps.

[0013] S1. Signal acquisition.

[0014] Real-time monitoring of the vibration acceleration and strain signals of the transformer core and windings.

[0015] S2. Modal decomposition.

[0016] The mixed vibration acceleration signal is separated into the core vibration acceleration signal and the winding vibration acceleration signal by wavelet packet transform. The core vibration acceleration signal is low frequency and the winding vibration acceleration signal is high frequency. The wavelet packet transform formula is: (3); where x(t) is the original vibration acceleration signal; i is the frequency band index; j is the decomposition level number; is the wavelet basis function; W i,j (t) is the coefficient of the i-th frequency band and the j-th layer decomposition after wavelet packet decomposition; the original vibration acceleration signal x(t) is decomposed into 5 layers using the db10 basis function to obtain the wavelet packet coefficient W of 32 frequency bands i,j (t); Calculate the energy proportion of each frequency band (4). Select the energy ratio E i The frequency band >85% is used as the control target, and the damping force target setting is set.

[0017] S3, damping force output.

[0018] The improved ADRC algorithm is used to dynamically adjust the magnetic field strength. The current regulation formula is: (5); where e(t) is the error signal, e(t)=F target -F actual ,k p 、k i 、k d is the ADRC parameter, K p =2.5; K i =0.8; K d =0.1; F target F is the target value / set value, which represents the target damping force for a specific channel, i.e. the low frequency of the main channel or the high frequency of the auxiliary channel; actual is the actual value / measured value, representing the actual damping force.

[0019] S4. Energy management.

[0020] The output voltage of the triboelectric nanogenerator (TENG) is stored in a supercapacitor after DC-DC conversion.

[0021] Furthermore, the collected vibration acceleration signal is preprocessed by Gaussian filtering. The preprocessing formula is: (6), where a filtered (t) is the acceleration signal after filtering; N is the number of samples involved in filtering; a(t k ) is at time point t k The original acceleration value measured at t is the current time point; t is the current time point; k is the kth sampling time point; σ is the standard deviation of the Gaussian filter.

[0022] Furthermore, the strain signal is compensated by using the temperature effect correction method. The strain signal compensation formula is: ε corrected =ε raw -α·ΔT(7);where ε corrected is the strain value after temperature compensation; ε raw is the original measured strain value; α is the temperature influence coefficient; ΔT is the temperature change.

[0023] Furthermore, for low-frequency vibration: target damping force , k1 is the proportional coefficient for low-frequency vibration, E low The frequency band identified as the main low-frequency vibration target, that is, the average energy value or total value of the low-frequency band in the frequency band with energy accounting for >85%; for high-frequency vibration: target damping force , k2 is the proportional coefficient for high frequency vibration, E high It is the frequency band identified as the main high-frequency vibration target, that is, the average energy value or total value of the high-frequency band in the frequency band with an energy share of >85%.

[0024] The beneficial effects of the present invention are: under 100Hz sinusoidal excitation, the vibration acceleration of the present invention is reduced by 82%. TENG outputs 7.3W of power under 1kHz vibration, and it only takes 30 minutes for the supercapacitor to charge to 12V. After 1000 hours of salt spray test, there is no corrosion on the cylinder surface, and the sealing ring leakage rate is <0.1mL / h. Through frequency band vibration suppression, environmental adaptive design and self-powered technology, the problems of insufficient low-frequency suppression, easy corrosion and high energy consumption of traditional vibration reduction devices are solved. Experiments show that the system vibration reduction efficiency is ≥80%, the energy self-sufficiency rate is 100%, and it is suitable for complex industrial environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is the overall architecture diagram of the transformer adaptive vibration reduction system based on magnetorheological damper of the present invention.

[0026] Figure 2 This is the structural diagram of the dual-channel magnetorheological damper.

[0027] Figure 3 This is a flow chart of the collaborative control method of the present invention. DETAILED DESCRIPTION

[0028] The transformer adaptive vibration reduction system based on a magnetorheological damper of the present invention includes a sensor module, a control unit, an actuator, and an energy recovery module. The transformer multimodal coordinated control method based on a magnetorheological damper of the present invention includes S1: signal acquisition, S2: modal decomposition, S3: damping force output, and S4: energy management. The structure and control method of the present invention are described in detail below with reference to the accompanying drawings.

[0029] like Figure 1 As shown, the transformer adaptive vibration reduction system based on magnetorheological damper includes a sensor module, a control unit, an actuator and an energy recovery module.

[0030] 1. Sensor module.

[0031] The sensor module includes an accelerometer and a fiber Bragg grating (FBG) strain sensor. The accelerometer is model PCB 352C33, with a bandwidth of 0.1-1kHz and a sensitivity of 10mV / g. The FBG strain sensor has an accuracy of ±0.1%, a resolution of 1με, and is resistant to electromagnetic interference. The accelerometer is used to monitor the vibration acceleration a(t) of the transformer core and windings in real time, while the FBG strain sensor is used to monitor the strain ε(t) of the transformer core and windings in real time.

[0032] Gaussian filtering is used to preprocess the vibration acceleration signal. The preprocessing formula is: (1).

[0033] Among them, a filtered (t) is the filtered acceleration signal, unit: m / s².

[0034] N is the number of samples involved in filtering.

[0035] a(t k ) is at time point t k The raw acceleration value measured at , in m / s².

[0036] t is the current time point, in seconds.

[0037] t k The kth sampling time point, unit: second.

[0038] σ is the standard deviation of Gaussian filtering, σ = 0.1 ms, or 0.1×10 −3 Second.

[0039] The strain signal is compensated by using the temperature effect correction method. The strain signal compensation formula is: ε corrected= ε raw -α·ΔT (2).

[0040] Among them, ε corrected It is the strain value after temperature compensation, unit: microstrain, i.e. με.

[0041] ε raw is the original measured strain value, unit: με.

[0042] α is the temperature influence coefficient, α=0.55με / ℃.

[0043] ΔT is the temperature change, unit: Celsius, that is, ℃.

[0044] II. Control unit.

[0045] The control unit is a real-time controller based on Xilinx Zynq-7000 FPGA, running an improved active disturbance rejection control ADRC algorithm, with a response time ≤1ms.

[0046] The ADRC control parameters include: (1) tracking differentiator TD parameters: tracking speed r, filter factor h0. (2) extended state observer ESO gain: β 01 , β 02 , β 03 , which determines the sensitivity and convergence speed of disturbance estimation. (3) nonlinear feedback NLSEF gain: proportional gain k p , differential gain k d . (4) frequency band control weight: damping force distribution ratio of low frequency and high frequency channels.

[0047] An improved active disturbance rejection control ADRC algorithm is adopted to convert the vibration acceleration signal into a magnetic field strength adjustment command in real time. Through active disturbance compensation and dynamic parameter optimization, the damping force is accurately matched to the target demand. The core structure of the ADRC controller includes the following.

[0048] (1) Tracking differentiator TD.

[0049] Generate smooth target signal v1(t) and its differential v2(t) to avoid command mutation and provide a smooth reference trajectory for ADRC. The algorithm formula of tracking differentiator is: (3).

[0050] Where v1(k) and v2(k) are the state variables of the tracking differentiator, unit: m / s².

[0051] h is the control step, unit: second.

[0052] Fhan is the fastest control synthesis function, used to generate the control signal.

[0053] F target is the target vibration signal, unit: m / s².

[0054] r and h0 are control parameters.

[0055] (2) Extended state observer ESO.

[0056] The total disturbance of the system, including external vibration interference, nonlinear characteristics of the magnetorheological fluid, time-varying of the damper parameters such as aging and temperature drift, is extended to the third state variable z3(k) of the system. The core function of the ESO is to estimate the total disturbance z3(k) in real time. By introducing the compensation term (z3(k) / b0), the controller can actively offset these disturbances, making the control amount u0 under the nominal model closer to the actual required control effect. In other words, z3(k) represents the total effect of all unmodeled and unmeasurable factors on the system behavior, and its estimation and compensation are the key to improving the robustness of the system. The algorithm formula of the extended state observer ESO is: (4).

[0057] Where z1(k), z2(k), and z3(k) are the state variables of the extended state observer, and the unit is related to vibration.

[0058] β 01 , β 02 , and β 03 are the observer gain parameters.

[0059] fal is a nonlinear error function used to suppress high-frequency noise.

[0060] e1(k) = Z1-F actual is the error between the observed damping force Z1 and the actual damping force F actual .

[0061] b0 is the control gain, representing the relationship between the unit current and the increase in damping force or equivalent magnetic field strength, with units such as N / A or T / A.

[0062] a1 is the power index parameter of the nonlinear error function fal, 0 < a1 < 1, which determines the strength of the nonlinear characteristics.

[0063] δ is the linear interval width parameter of the nonlinear error function fal, with units of observation error, such as m / s², used to control the segmentation range of the function.

[0064] u(k) is the control input signal, i.e., the magnetic field coil current command value at time k, with units of A.

[0065] (3) Nonlinear feedback NLSEF.

[0066] Combining the nonlinear characteristics of the error, the control amount u0 is dynamically adjusted, and the calculation formula is: (5).

[0067] Where k p is the proportional gain, and k d is the derivative gain, optimized by the particle swarm algorithm PSO.

[0068] α1, α2, α3, and δ are nonlinear function parameters.

[0069] Final control amount is the current command value that directly drives the field coil. Z3 is the total disturbance estimated by the extended state observer (ESO). In the ADRC framework, Z3 captures the sum of all unmodeled dynamics, external disturbances, and internal parameter changes, such as magnetorheological fluid aging, temperature drift, and nonlinearity, that affect the system. It represents all deviations between the actual system and the nominal model within the controller.

[0070] By estimating Z3 in real time, ESO allows the controller to actively compensate for these disturbances, significantly improving the system’s robustness to external variations and uncertainties.

[0071] The roles of ADRC in this invention include: (1) Active disturbance compensation, which uses ESO to estimate the total disturbance in real time, including magnetorheological fluid nonlinearity and external vibration interference, significantly improving robustness. (2) Frequency band coordination, which designs a dual-channel control strategy for low-frequency core and high-frequency winding vibration to achieve broadband suppression. (3) Self-powered support, in which the energy recovered by TENG is directly used in the ADRC controller, reducing external dependence. (4) Parameter self-optimization, in which the PSO algorithm dynamically adjusts the control parameters to adapt to complex working conditions. In the transformer vibration reduction system based on the improved ADRC, parameter optimization and closed-loop feedback are the key links to achieve adaptive control.

[0072] The following control algorithms are built into the control unit.

[0073] Algorithm 1, vibration mode decomposition.

[0074] The mixed vibration signal is separated into the core vibration signal and the winding vibration signal by wavelet packet transform. The core vibration signal is low frequency and its frequency is in the range of 50Hz-200Hz; the winding vibration signal is high frequency and its frequency is in the range of 200Hz-1KHz.

[0075] The wavelet packet transform formula is: (6). Where x(t) is the original vibration signal; i is the frequency band index; j is the number of decomposition levels; is the wavelet basis function; W i,j (t) is the coefficient of the i-th frequency band and j-th layer decomposition after wavelet packet decomposition. The original vibration signal x(t) is decomposed into 5 layers using the db10 basis function to obtain the wavelet packet coefficients W of 32 frequency bands. i,j (t).

[0076] According to the wavelet packet coefficient W i,j (t) Calculate the energy proportion of each frequency band (7). Select the energy ratio The frequency band >85% is used as the control target, and the damping force target setting is set.

[0077] For low frequency vibrations: Target damping force k1 is the proportional coefficient for low frequency vibration. E low This is the average or total energy value of the low-frequency band identified as the primary low-frequency vibration target, i.e., the frequency band with an energy share greater than 85%. Low frequency refers to 50Hz-200Hz.

[0078] For high frequency vibrations: Target damping force k2 is the proportional coefficient for high frequency vibration. high This is the average or total energy value of the high-frequency band in the frequency band identified as the primary high-frequency vibration target, that is, the energy proportion of the frequency band greater than 85%. High frequency refers to 200Hz-1kHz.

[0079] The basis function of wavelet packet transform is db10, and the number of decomposition layers is 5.

[0080] Algorithm 2: Adaptive damping force adjustment.

[0081] The improved ADRC algorithm is used to dynamically adjust the magnetic field strength. The current regulation formula is: (8). Where, e(t) is the error signal, e(t)=F target -F actual ,k p 、k i 、k d is the ADRC parameter, and K is obtained by particle swarm optimization. p =2.5; K i =0.8; K d =0.1. F target F is the target value / set value, which represents the target damping force for a specific channel, that is, the low frequency of the main channel or the high frequency of the auxiliary channel, in Newtons; actual It is the actual value / measured value, representing the actual damping force in Newton.

[0082] Optimize the parameter k by particle swarm optimization p 、k i 、k d , β 01 , β 02 and β 03 , to minimize the vibration energy integral In order to optimize the indicators, the particle swarm algorithm is used for optimization processing, and the optimization processing method includes the following steps.

[0083] S1. Initialization.

[0084] Randomly generate a set of parameter combinations, such as kp , k d , β 01 .

[0085] S2, evaluate fitness.

[0086] Substitute each set of parameters into the ADRC controller, simulate or experimentally measure the vibration suppression effect, and calculate the minimum vibration energy integral J.

[0087] S3, update particles.

[0088] According to the individual historical optimal position p best and the global optimal position g best , adjust the parameter direction.

[0089] The particle update formula is: (9).

[0090] x i k is the position vector of particle i at the kth iteration, representing a set of parameters.

[0091] v i k is the velocity vector of particle i at the kth iteration.

[0092] p best is the individual historical optimal position of particle i up to the kth iteration.

[0093] g best is the global optimal position of the entire population up to the kth iteration.

[0094] c1, c2 are acceleration constants, commonly known as individual cognitive factor and social cognitive factor, for example, c1 = c2 = 2.0.

[0095] r1, r2 are uniformly distributed random numbers in the interval [0, 1].

[0096] w is the inertia weight, which controls the speed inheritance, for example, its value decreases linearly from 0.9 to 0.4.

[0097] S4, iteration convergence.

[0098] Repeat steps S2-S3 until the minimum vibration energy integral J reaches a minimum value or the number of iterations is exhausted.

[0099] The optimization process is a frequency band cooperative control. The low frequency channel uses PID regulation to ensure stability, and the high frequency channel introduces a nonlinear compensation fal function to enhance anti-interference ability.

[0100] Optimization results: the optimal parameter combination is obtained, so that ADRC can quickly respond and stably control under different working conditions, such as low-frequency vibration, high-frequency disturbance and salt spray corrosion.

[0101] The feedback signal source of the control unit is two, one is the acceleration sensor, which measures the vibration acceleration a(t) in real time, reflecting the actual effect of damping force. The second is the fiber Bragg grating strain sensor, which monitors the state of the magnetorheological fluid, such as liquid level and temperature drift.

[0102] After the signal is filtered by Gaussian filter to eliminate noise and temperature compensation, error calculation is performed, that is, the target damping force is compared with the actual damping force, and the error signal e(t) is generated. If e(t) is continuously large, the PSO algorithm is triggered to re-optimize the parameters. The optimized parameters are loaded to the ADRC controller in real time. Through continuous feedback, the system can cope with the uncertainty of magnetorheological fluid aging and external load mutation. For different frequency bands of vibration, the main / supplementary channel parameters are adjusted independently to avoid low-frequency over-damping or high-frequency under-damping.

[0103] Three, actuator.

[0104] The actuator includes a double-channel magnetorheological damper, the main channel has a spiral flow channel with a diameter of 8 mm, and the supplementary channel has a rotary valve with an opening degree of 0-90° and adjustable.

[0105] The main channel is used for low-frequency suppression, and the spiral flow channel design is adopted, the Reynolds number Re<2000, and the pressure drop ΔP≤0.2 MPa.

[0106] The supplementary channel is used for high-frequency suppression, and the rotary valve adjusts the liquid flow turbulence intensity to suppress vibration above 200 Hz.

[0107] According to the current command u(t) output by the ADRC, the magnetic field intensity is adjusted through the coil current to output the double-channel damping force. The main channel is used for low-frequency suppression of 50-200 Hz, and the supplementary channel is used for high-frequency suppression of 200-1 kHz. The magnetic field intensity B(t) of the main channel and the supplementary channel is controlled by the current: (10).

[0108] Wherein, B(t) is the magnetic field intensity, unit: Tesla, namely T.

[0109] μ0 is the magnetic permeability of vacuum, μ0=4π×10 −7 H / m.

[0110] N is the number of turns of the coil, N=200.

[0111] u(t) is the current, unit: ampere, namely A.

[0112] l is the length of the magnetic circuit, unit: meter, l=0.1 m.

[0113] Four, energy recovery module.

[0114] The energy recovery module comprises a triboelectric nanogenerator TENG and a super capacitor, the triboelectric nanogenerator TENG is embedded in a damper piston rod, a multi-layer triboelectric structure PTFE-Cu composite film is adopted, and the output power is 5W-8W. The TENG embedded in the damper piston rod outputs and is stored in the super capacitor after being stabilized by a Buck-Boost circuit, so that energy closed loop utilization is realized.

[0115] The output voltage of the triboelectric nanogenerator TENG is: (11)。

[0116] Wherein, V OC is the open circuit voltage, unit: volt. V OC =1200V.

[0117] σ is the surface charge density, unit: microcoulomb per square meter. σ=10μC / m 2 .

[0118] d is the distance between the friction layers, unit: microns. d=100μm.

[0119] ε0 is the vacuum permittivity, ε0=8.85×10 −12 F / m.

[0120] The regulation formula of the Buck-Boost circuit duty cycle D is: D=V out / (V in + V out ), wherein V in is the input voltage, V in =12V. V out is the output voltage, V out =5V.

[0121] The triboelectric nanogenerator TENG converts vibration mechanical energy into electrical energy, and the conversion efficiency is ≥65%, the output voltage of the triboelectric nanogenerator TENG is stored in the super capacitor after DC-DC conversion, the capacity of the super capacitor is 100F, the cycle life is ≥10 6 times, and full self-power supply of the control unit is realized. The energy management equation is: (12)。

[0122] Wherein, E cap is the energy stored in the super capacitor, unit: joule, i.e. J.

[0123] C is the capacitance value, C=100F.

[0124] V cap is the capacitor voltage, unit: volt, V cap ≤12V.

[0125] According to the use environment, the present application is designed as follows.

[0126] (1) Corrosion-resistant magnetorheological fluid design.

[0127] Corrosion-resistant magnetorheological fluid: Fe3O4SiO2core-shell nanoparticles with a particle size of 50-100 nm and a saturation magnetization of ≥80 emu / g are used, and the antioxidant performance is improved by 3 times in a salt spray environment.

[0128] (2) Sealing and corrosion-resistant structure.

[0129] The surface of the cylinder is a 16L stainless steel substrate that has been sandblasted and then coated with a PTFE coating through CVD process, with a coating thickness of 50μm and an adhesion test standard ASTM D3359 of ≥4B.

[0130] The sealing ring is made of fluororubber, which can withstand temperatures of -40°C to 200°C, and passes the ISO 9227 salt spray test for 1000 hours without corrosion. Its compression formula is , the original thickness of the sealing ring t0=5mm, the compressed thickness t c =3.5mm, and the compression rate =30%.

[0131] The life prediction model Arrhenius equation is (13).

[0132] L is the failure time of the sealing ring or magnetorheological fluid predicted under the accelerated temperature T condition.

[0133] L0 is the reference failure time of the sealing ring or magnetorheological fluid at the reference temperature T0.

[0134] E a is the activation energy of the sealing material, which represents the sensitivity of the material aging reaction to temperature.

[0135] R is the universal gas constant.

[0136] T is the temperature at which the accelerated aging test is performed.

[0137] T0 is the reference temperature at which the sealing ring or magnetorheological fluid is actually working or stored.

[0138] (3) Anti-settling module.

[0139] Ni-Mn-Ga shape memory alloy springs are used, with a phase transition temperature T trans =50 degrees Celsius and a strain recovery rate of ≥95%; its frequency control equation is (14).

[0140] The application further provides a transformer adaptive damping multi-modal collaborative control method based on a magneto-rheological damper.

[0141] S1, signal acquisition.

[0142] Real-time monitoring of vibration acceleration and strain signals of the transformer core and winding.

[0143] S2, modal decomposition.

[0144] The mixed vibration acceleration signal is separated into core vibration acceleration signals and winding vibration acceleration signals through wavelet packet transformation, wherein the core vibration acceleration signals are low frequency, and the winding vibration acceleration signals are high frequency; the wavelet packet transformation formula is: (15); wherein x(t) is the original vibration acceleration signal; i is the frequency band index; j is the decomposition layer number; is the wavelet base function; W i,j (t) is the coefficient of the i-th frequency band and the j-th layer decomposition after wavelet packet decomposition; the original vibration acceleration signal x(t) is decomposed by using db10 base function for 5 layers to obtain 32 frequency band wavelet packet coefficients W i,j (t); the energy proportion of each frequency band is calculated (16). The frequency band with an energy proportion E i >85% is selected as the control target, and the damping force target setting is set.

[0145] S3, damping force output.

[0146] The improved ADRC algorithm is used to dynamically adjust the magnetic field strength, and the current regulation formula is: (17); wherein e(t) is an error signal, e(t)=F target -F actual , k p , k i , k d are ADRC parameters, K p =2.5; K i =0.8; K d =0.1; F target is a target value / set value, representing the target damping force for a specific channel, i.e., the main channel low frequency or the auxiliary channel high frequency; F actual is an actual value / measured value, representing the actual damping force.

[0147] S4, energy management.

[0148] The output voltage of the friction nanogenerator TENG is stored in the super capacitor after DC-DC conversion.

[0149] The installation and maintenance of the application are described below.

[0150] (1) System installation.

[0151] The double-channel magneto-rheological damper is symmetrically installed at four corners of the transformer base, and the acceleration sensor and the fiber Bragg strain sensor of the sensor module are arranged on the core and the winding.

[0152] (2) Debugging process.

[0153] The ADRC parameters are initialized through the LabVIEW platform of the upper computer software, white noise excitation test is carried out, and the control bandwidth is optimized.

[0154] (3) Maintenance measures.

[0155] The magneto-rheological fluid viscosity is checked every six months, the target value of the magneto-rheological fluid viscosity is 0.8-1.2 Pa·s, the service life of the fluororubber sealing ring is five years, and the fluororubber sealing ring is replaced every five years.

[0156] (4) Actual use.

[0157] The transformer runs in a salt spray environment, the viscosity of the magneto-rheological fluid decreases due to corrosion, and the damping force is insufficient.

[0158] Closed-loop feedback process: ① sensor detection: the acceleration sensor finds that the vibration amplitude increases, and the fiber Bragg strain sensor detects that the liquid level decreases. ② error generation: F actual <F target , the error e(t) increases. ③ parameter optimization: the PSO algorithm re-adjusts the ESO gain β 02 , improves the disturbance estimation speed of the liquid level change. Increase k p of NLSEF, enhance the proportional control effect, and compensate for the loss of damping force. ④ execution effect: the magnetic field strength command B(t) is improved, the yield stress τ y of the magneto-rheological fluid is restored, and the vibration suppression effect is improved.

[0159] The vibration system and the multi-modal collaborative control method of the application have the following advantages.

[0160] 1. Damping effect.

[0161] Under the excitation of a 100Hz sine wave, the vibration acceleration of the application is reduced by 82%. However, under the excitation of a sine wave with the same frequency, the vibration acceleration of the traditional rubber pad is only reduced by 45%.

[0162] 2. Energy recovery.

[0163] The TENG outputs a power of 7.3W under a 1kHz vibration, and it only takes 30 minutes to charge a super capacitor to 12V.

[0164] 3. Environmental testing.

[0165] After 1000 hours of salt spray test, the cylinder surface has no corrosion, and the leakage rate of the sealing ring is less than 0.1 mL / h.

[0166] The application solves the problems of low-frequency suppression deficiency, easy corrosion and high energy consumption of the traditional damping device through frequency band vibration suppression, environment adaptive design and self-power supply technology. Experiments show that the system damping efficiency is greater than or equal to 80%, the energy self-sufficiency rate is 100%, and the system is suitable for complex industrial environments.

Claims

1. A transformer adaptive vibration reduction system based on magnetorheological dampers, characterized in that: It includes a sensor module, a control unit, an actuator and an energy recovery module; the sensor module is used to monitor the vibration acceleration and strain of the transformer core and winding in real time; the control unit converts the vibration acceleration signal into a magnetic field strength adjustment instruction in real time, and through active compensation of disturbances and dynamic optimization of parameters, the damping force is accurately matched to the target requirements; the actuator includes a dual-channel magnetorheological damper, the main channel performs low-frequency suppression, and the auxiliary channel performs high-frequency suppression; the energy recovery module includes a friction nanogenerator TENG and a supercapacitor. The friction nanogenerator TENG is embedded in the damper piston rod and adopts a multi-layer triboelectric structure PTFE-Cu composite film. The output voltage of the friction nanogenerator TENG is stored in the supercapacitor after DC-DC conversion, realizing closed-loop energy utilization.

2. The transformer adaptive vibration reduction system based on magnetorheological damper according to claim 1 is characterized in that: The sensor module includes an acceleration sensor and a fiber grating strain sensor.

3. The transformer adaptive vibration reduction system based on magnetorheological damper according to claim 2, characterized in that: The control unit includes a tracking differentiator TD, an extended state observer ESO and a nonlinear feedback NLSEF.

4. The transformer adaptive vibration reduction system based on magnetorheological damper according to claim 3 is characterized in that: The main channel adopts a spiral flow channel design with a spiral flow channel diameter of 8mm, a Reynolds number Re<2000, and a pressure drop ΔP≤0.2MPa.

5. The transformer adaptive vibration reduction system based on magnetorheological damper according to claim 4 is characterized in that: The opening of the rotary valve of the auxiliary channel is 0-90°. The rotary valve is used to adjust the turbulence intensity of the liquid flow and suppress vibrations above 200Hz.

6. The transformer adaptive vibration reduction system based on magnetorheological damper according to claim 5, characterized in that: The tracking differentiator TD generates a smooth target signal v1(t) and its differential v2(t). The algorithm formula of the tracking differentiator is: (1); Where v1(k) and v2(k) are the state variables of the tracking differentiator; h is the control step size; Fhan is the fastest control synthesis function used to generate the control signal; F target is the target vibration signal; r and h0 are control parameters.

7. The transformer adaptive vibration reduction system based on magnetorheological damper according to claim 6, characterized in that: The extended state observer (ESO) expands the total system disturbance, including external vibration interference, nonlinear characteristics of the magnetorheological fluid, and time-varying damper parameters, into the third state variable z3(k) of the system. The algorithm formula of the extended state observer (ESO) is: (2); where z1(k), z2(k), and z3(k) are the state variables of the extended state observer; β 01 , β 02 , β 03 is the observer gain parameter; fal is the nonlinear error function used to suppress high-frequency noise; e1(k)=Z1-F actual , is the observed value Z1 of the damping force and the actual value F of the damping force actual b0 is the control gain, which characterizes the influence of unit current on the increment of damping force or equivalent magnetic field intensity; a1 is the power exponential parameter of the nonlinear error function fal, which determines the intensity of the nonlinear characteristic; δ is the linear interval width parameter of the nonlinear error function fal, which is used to control the segmented range of the function; u(k) is the control input signal, that is, the magnetic field coil current command value output at time k.

8. The multi-mode cooperative control method of the transformer adaptive vibration reduction system based on magnetorheological dampers according to claim 7 is characterized in that: The following steps are involved: S1. Signal acquisition Real-time monitoring of the vibration acceleration and strain signals of the transformer core and windings; S2. Modal decomposition The mixed vibration acceleration signal is separated into the core vibration acceleration signal and the winding vibration acceleration signal by wavelet packet transform. The core vibration acceleration signal is low frequency and the winding vibration acceleration signal is high frequency. The wavelet packet transform formula is: (3); where x(t) is the original vibration acceleration signal; i is the frequency band index; j is the decomposition level number; is the wavelet basis function; W i,j (t) is the coefficient of the i-th frequency band and the j-th layer decomposition after wavelet packet decomposition; the original vibration acceleration signal x(t) is decomposed into 5 layers using the db10 basis function to obtain the wavelet packet coefficient W of 32 frequency bands i,j (t); Calculate the energy proportion of each frequency band (4) Select the energy ratio E i The frequency band >85% is used as the control target, and the damping force target setting is set; S3, damping force output The improved ADRC algorithm is used to dynamically adjust the magnetic field strength. The current regulation formula is: (5); where e(t) is the error signal, e(t)=F target -F actual ,k p 、k i 、k d is the ADRC parameter, K p =2.5; K i =0.8; K d =0.1; F target F is the target value / set value, which represents the target damping force for a specific channel, i.e. the low frequency of the main channel or the high frequency of the auxiliary channel; actual is the actual value / measured value, representing the actual damping force; S4. Energy Management The output voltage of the triboelectric nanogenerator (TENG) is stored in a supercapacitor after DC-DC conversion.

9. The transformer adaptive vibration reduction multi-mode coordinated control method based on magnetorheological damper according to claim 8 is characterized in that: The collected vibration acceleration signal is preprocessed by Gaussian filtering. The preprocessing formula is: (6), where a filtered (t) is the acceleration signal after filtering; N is the number of samples involved in filtering; a(t k ) is at time point t k The original acceleration value measured at t is the current time point; t is the current time point; k is the kth sampling time point; σ is the standard deviation of the Gaussian filter.

10. The transformer adaptive vibration reduction multi-mode coordinated control method based on magnetorheological damper according to claim 9 is characterized in that: The strain signal is compensated by using the temperature effect correction method. The strain signal compensation formula is: ε corrected =ε raw -α·ΔT (7); where ε corrected is the strain value after temperature compensation; ε raw is the original measured strain value; α is the temperature influence coefficient; ΔT is the temperature change.

11. The transformer adaptive vibration reduction multi-mode coordinated control method based on magnetorheological damper according to claim 10 is characterized in that: For low frequency vibrations: Target damping force , k1 is the proportional coefficient for low-frequency vibration, E low The frequency band identified as the main low-frequency vibration target, that is, the average energy value or total value of the low-frequency band in the frequency band with energy accounting for >85%; for high-frequency vibration: target damping force , k2 is the proportional coefficient for high frequency vibration, E high It is the frequency band identified as the main high-frequency vibration target, that is, the average energy value or total value of the high-frequency band in the frequency band with an energy share of >85%.

Images