Methods and systems for controlling the pressure of core stacking to reduce transformer no-load losses

Abstract

This invention discloses a method and system for controlling the pressure of stacked iron cores to reduce transformer no-load losses. The method includes acquiring the excitation current signal of the stacked iron cores under constant-voltage sinusoidal excitation conditions; extracting the harmonic characteristic components of the signal and decoupling them into stress characterization indicators and air gap characterization indicators, calculating the corresponding hysteresis loss components and air gap additional loss components; gradually increasing the clamping force under a monotonically increasing pressure application path, determining the total loss gradient based on the components at each pressure step, and locating the first optimal pressure point; acquiring pre-configured compensation parameters, performing offset compensation calculations on the first optimal pressure point, obtaining the target curing pressure, and locking it. This invention, through electrical harmonic analysis, can solve the deviations caused by magnetic coupling hysteresis and process relaxation, achieving optimized control of no-load losses.

Description

Technical Field

[0001] This invention relates to the field of transformer manufacturing and assembly control technology, and in particular to a method and system for controlling the pressure of core stacking to reduce transformer no-load loss. Background Technology

[0002] As one of the main components for electromagnetic energy conversion, the quality of the transformer core determines the transformer's no-load loss level. In the manufacturing process of large power transformers, the core is usually made of multi-stage stepped lapped grain-oriented silicon steel sheets.

[0003] Currently, the industry typically relies on mechanical strain gauges or piezoelectric sensors for online monitoring of physical forces when controlling the stacking pressure of iron cores, and sets the curing pressure target based on empirical thresholds. Regarding optimized control strategies, some researchers have proposed various feedback adjustment schemes based on measured losses. For example, a temporary test winding is wound onto the iron core, and the total no-load loss is measured in real time during pressurization. When the total loss decrease curve flattens out or shows an upward trend, pressurization is stopped, and curing is performed in this state. Another approach uses an iterative trial-and-error algorithm with repeated pressure increases and decreases to find the point of lowest loss.

[0004] Existing control schemes cannot effectively suppress complex measurement interference when facing complex magnetic coupling systems, and it is also difficult to build a high-precision adaptive control mechanism under irreversible conditions, making it difficult for the iron core to lock into the optimal working state. Summary of the Invention

[0005] Purpose of the invention: To provide a method and system for controlling the pressure of stacked iron cores to reduce the no-load loss of transformers, in order to solve the above-mentioned technical problems existing in the prior art.

[0006] Technical Solution: First Aspect: A method for controlling the pressure of core stacking to reduce transformer no-load losses, comprising:

[0007] Obtain the excitation current signal of the stacked iron core under constant voltage sinusoidal excitation conditions;

[0008] Harmonic characteristic components of the excitation current signal are extracted, and the harmonic characteristic components are decoupled into stress characterization index and air gap characterization index. The corresponding hysteresis loss component and air gap additional loss component are calculated.

[0009] Under the monotonically increasing pressure application path, the clamping force of the stacked iron core is gradually increased. Acquisition and extraction, decoupling and calculation are performed in each pressure step. Based on the hysteresis loss component and the air gap additional loss component in each pressure step, the total loss gradient is determined, and the first optimal pressure point on the rising branch is located according to the total loss gradient.

[0010] Obtain the compensation parameters, and perform offset compensation calculation on the first optimal pressure point based on the compensation parameters to obtain the target solidification pressure;

[0011] The stacked iron cores are locked according to the target curing pressure.

[0012] The second aspect: a core stacking pressure control system for reducing transformer no-load losses, the system including a processor, the processor being coupled to a memory, the memory storing instructions, the instructions being executed by the processor to implement the method in any possible implementation of the first aspect.

[0013] Beneficial effects: This invention solves the deviations caused by magnetic coupling hysteresis and process relaxation through electrical harmonic analysis, and achieves optimized control of no-load loss. Attached Figure Description

[0014] Figure 1 This is a flowchart of the core stacking pressure control method for reducing transformer no-load loss according to the present invention.

[0015] Figure 2 This is a flowchart illustrating the decoupling of harmonic characteristic components according to the present invention.

[0016] Figure 3 This is a flowchart for determining the total loss gradient according to the present invention.

[0017] Figure 4 This is a flowchart illustrating the locking process for stacked iron cores according to the present invention.

[0018] Figure 5 This is a flowchart of the offline construction process for obtaining pre-configured parameters according to the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0020] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein are implemented in a sequence other than those illustrated or described herein. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those explicitly listed, but may include other steps or units not explicitly listed or inherent to the process, method, product, or apparatus.

[0021] Based on the problems mentioned in the background technology, the current control scheme still has several problems in complex magnetic coupling systems. For example, the total loss is measured at a single point, making it difficult to distinguish the independent influence weights of mechanical force on the joint air gap and material magnetic domains, resulting in a lack of state assessment basis when looking for the equilibrium point. Another example is that there is a historical dependence and irreversibility between the applied pressure and the magnetic performance response. Small fluctuations in the pressurization process and solidification decay can easily cause the final working state to deviate from the theoretical optimal value at the time of control.

[0022] In this application, "stacked core" refers to the transformer magnetic circuit body formed by stacking oriented silicon steel sheets in a predetermined overlapping structure; "constant voltage sinusoidal excitation condition" refers to applying a sinusoidal excitation voltage of rated amplitude to the test winding of the stacked core through an AC power supply with low output impedance.

[0023] According to one aspect of this application, this embodiment provides an adaptive control method for core stacking pressure to reduce transformer no-load losses, the method comprising:

[0024] Step 101: Obtain the excitation current signal of the stacked iron core under constant voltage sinusoidal excitation conditions.

[0025] Specifically, the stacked iron cores are assembled into an adjustable clamping mechanism, and an initial preload F is applied to the stacked iron cores by a servo electric actuator. _0 Initial preload F _0 The setting can be set to the lower limit of the allowable clamping force range for the corresponding silicon steel. For example, for grain-oriented silicon steel of grade 30ZH100, the initial preload F... _0 It can be set to 5kN.

[0026] To ensure the purity of the excitation voltage waveform, the short-circuit impedance of the AC power supply should be less than 1% of the excitation impedance of the stacked iron core. Based on this setting, the voltage drop across the power supply impedance caused by harmonic components in the excitation current can be ignored. On this basis, the peak magnetic flux density B of the stacked iron core... _mThe excitation current signal is determined by the applied voltage. In this embodiment, the excitation current waveform i(t) is synchronously acquired at a sampling rate of not less than 20 times the fundamental frequency to obtain the excitation current signal.

[0027] Step 102: Extract the harmonic characteristic components of the excitation current signal, decouple the harmonic characteristic components into stress characterization index and air gap characterization index, and calculate the corresponding hysteresis loss component and air gap additional loss component. That is, perform discrete Fourier decomposition on the excitation current signal to extract the fundamental amplitude and at least one odd-order higher harmonic amplitude; use the fundamental amplitude and the odd-order higher harmonic amplitude together as the harmonic characteristic components.

[0028] In this embodiment, the acquired excitation current signal is decomposed using Discrete Fourier Transform to extract the fundamental amplitude I. _1 and at least one odd-order higher harmonic amplitude, such as the third harmonic amplitude I. _3 and the amplitude of the fifth harmonic I _5 Together, they constitute the harmonic characteristic components.

[0029] The physical model of the decoupling logic is as follows:

[0030] N×i(t)=H(B(t),σ)×l _c +B(t)×g / μ _0 ;

[0031] Where N is the number of turns in the winding, i(t) is the excitation current signal, H is the nonlinear magnetization function of the core material under compressive stress σ, and l _c Let B(t) be the effective magnetic path length, B(t) be the magnetic flux density as a function of time, g be the equivalent total air gap length, and μ be the effective magnetic path length. _0 is the vacuum permeability.

[0032] Specifically, since the air gap component only contributes to the fundamental magnetic force and does not generate higher harmonics, the extracted third harmonic amplitude I _3 This is the stress characterization index.

[0033] Furthermore, the current equivalent compressive stress σ is determined using a pre-calibrated stress-harmonic relationship function h(σ). At the fundamental amplitude I... _1 The fundamental magnetization component of the material at the corresponding stress level is subtracted to obtain the air gap characterization index. Using a stress-corrected loss model, the hysteresis loss component is calculated. Combined with the air gap loss coefficient and the air gap characterization index, the additional air gap loss component is then calculated.

[0034] Step 103: Under the monotonically increasing pressure application path, the clamping force of the stacked iron core is gradually increased. Acquisition and extraction, decoupling and calculation are performed under each pressure step. Based on the hysteresis loss component and the air gap additional loss component under each pressure step, the total loss gradient is determined, and the first optimal pressure point on the rising branch is located according to the total loss gradient.

[0035] Because the magnetic domain structure inside grain-oriented silicon steel exhibits irreversible pinning under stress, repeated pressure adjustments can easily cause the measurement data to enter different hysteresis branches. Therefore, this embodiment sets the pressure search as a monotonically increasing sequence. Specifically, the pressure step increment Δ is set... _F Each additional pressure step increment Δ _F After the pressure sensor readings stabilize, the data acquisition and decoupling process is repeated. The corresponding calculation formula is:

[0036] Gal=(P _hyst_j -P _hyst_j-1 ) / Δ _F +(P _gap_j -P _gap_j-1 ) / Δ _F ;

[0037] Among them, P _hyst_j P represents the hysteresis loss component under the current pressure step. _hyst_j-1 P represents the hysteresis loss component under the previous pressure step. _gap_j Add a loss component to the air gap at the current pressure step, P _gap_j-1 This is an additional loss component in the air gap from the previous pressure step.

[0038] As the pressure increases progressively, the total loss gradient monotonically increases from a negative value to a positive value. When the gradient values ​​of two adjacent steps reverse, linear interpolation is used to locate the position where the total loss gradient is zero. The pressure value corresponding to this position is the first optimal pressure point F. _star .

[0039] Step 104: Obtain compensation parameters, and perform offset compensation calculation on the first optimal pressure point based on the compensation parameters to obtain the target curing pressure.

[0040] Specifically, the compensation parameters are pre-configured parameters, including the hysteresis asymmetric parameter η. _mm and pressure relaxation parameter Δ _relax The offset compensation calculation formula is:

[0041] Δ _F_corr =(1-η _mm )×G _hyst_star +η _mm ×G _hyst_prime ×Δ _relax +G _gap_prime ×Δ_relax ) / (η _mm ×G _hyst_prime +G _gap_prime );

[0042] Where, Δ _F_corr For offset compensation, G _hyst_star G represents the hysteresis loss gradient at the first optimal pressure point. _hyst_prime G is the rate of change of the hysteresis loss gradient with pressure. _gap_prime Let F be the rate of change of the air gap loss gradient with pressure. The first optimal pressure point F... _star With offset compensation amount Δ _F_corr Adding them together, we get the target curing pressure F. _opt This correction ensures that the actual effective pressure of the stacked iron core, after curing and relaxation, converges within the neighborhood of the minimum loss value. In this embodiment, since the first optimal pressure point is determined during the pressurization process, and the stacked iron core is prone to pressure relaxation during the subsequent curing process, thus switching to the pressure reduction branch operation, offset compensation is required.

[0043] Step 105: Lock the stacked iron core according to the target curing pressure.

[0044] After determining the target curing pressure, the output of the clamping mechanism is adjusted by controlling the servo electric actuator to ensure that the real-time pressure sensor reading reaches the target curing pressure value. A mechanical locking command is then executed, or the heating curing program is initiated. This adaptive adjustment can offset the material's magnetic coupling hysteresis effect and mechanical relaxation deviations during production. In this adaptive control method, the deviation between the final curing operating point and the theoretical minimum loss value of the pressure reduction branch is jointly determined by the search step and the accuracy of the compensation model, achieving approximation within engineering limits. Compared to fixed empirical thresholds, this method can adaptively determine the optimal pressure for each core, avoiding deviations caused by material property fluctuations between different batches and achieving global optimization of no-load losses.

[0045] In one possible implementation, the harmonic characteristic components are decoupled into stress characterization indicators and air gap characterization indicators, specifically including:

[0046] Step 201: The third harmonic amplitude among the odd-order higher harmonic amplitudes is used as a stress characterization index. Based on the single-piece stress harmonic calibration function, the third harmonic amplitude is mapped to the equivalent compressive stress.

[0047] Specifically, based on the harmonic air gap invariance under constant voltage sinusoidal excitation conditions, under sinusoidal voltage excitation and low output impedance of the excitation power supply, the magnetic flux density waveform within the core is locked to a fundamental sinusoidal wave by the applied voltage. Correspondingly, the equivalent air gap at the lamination joints only acts on the fundamental magnetic force and does not participate in the generation of any higher harmonics.

[0048] The amplitude of the third and higher odd harmonics in the excitation current is determined by the nonlinear magnetization characteristics of the grain-oriented silicon steel material. The amplitude of these odd harmonics is sensitive to compressive stress and exhibits a monotonically increasing trend.

[0049] Based on this, the third harmonic amplitude is extracted from the odd-order higher harmonic amplitudes to characterize the stress state. Since the stress distribution on the actual core cross-section may be non-uniform, the equivalent compressive stress is calculated through mapping. This equivalent compressive stress can be expressed as an equivalent scalar, ensuring that a uniformly stressed core produces the same third harmonic amplitude as an actual non-uniformly stressed core. This equivalent scalar characterizes the weighted average stress state. The mapping process in this embodiment is expressed as follows:

[0050] σ=h _inv_3 (I _3 );

[0051] Where σ is the equivalent compressive stress, I _3 For the extracted third harmonic amplitude, h _inv_3 This is the inverse function of the pre-configured monolithic stress harmonic calibration function. The monolithic stress harmonic calibration function is a reference curve established by pre-measuring silicon steel sheets from the same batch on a gapless monolithic magnetic tester.

[0052] Step 202: Obtain the corresponding pre-stored gapless magnetization fundamental wave component based on the equivalent compressive stress;

[0053] The measured fundamental amplitude is a mixed signal, including the current required for the material's own magnetization and the current required to overcome the air gap. To separate the two types of current, based on equivalent compressive stress, the corresponding pre-stored air-gap-free magnetized fundamental component is retrieved from a pre-constructed data dictionary or fitted curve. This component characterizes the magnitude of the fundamental current required to drive the core to its rated magnetic flux density under the current stress level, assuming the core is ideally closed and completely without a seam air gap.

[0054] Step 203: Subtract the fundamental wave component without air gap magnetization from the fundamental wave amplitude to obtain the air gap characterization index, that is, obtain the air gap characterization index with the influence of the magnetization of the isolation material.

[0055] After obtaining the fundamental component of the air-gap-free magnetization, a stripping operation is performed, which is expressed as follows:

[0056] ξ _g =I _1 -I _1_core (σ);

[0057] Where, ξ _g I is an indicator for air gap characterization. _1 I represents the measured fundamental amplitude. _1_core(σ) represents the pre-stored air-gap-free magnetized fundamental component under the corresponding equivalent compressive stress. The difference obtained by subtraction is the additional fundamental excitation current required to drive the magnetic flux through all the air gaps in the lamination joints. Through the decoupling process, the mixed stress effect and the air gap effect can be transformed into two independent indicators, avoiding the interference of air gap changes on stress monitoring.

[0058] Since high-order harmonics all exhibit air gap invariance, to avoid single data failures caused by local saturation, sensor drift, or extremely non-uniform stress distribution, this embodiment also provides an alternative implementation method including cross-consistency verification, specifically including:

[0059] Step 204: The fifth harmonic amplitude among the odd-order higher harmonic amplitudes is mapped to the verification compressive stress through the corresponding pre-configured single-piece stress harmonic calibration function.

[0060] Using the extracted fifth harmonic amplitude as an independent redundant verification source, its operational logic is expressed as follows:

[0061] σ _5 =h _inv_5 (I _5 );

[0062] Where, σ _5 To verify the compressive stress, I _5 For the extracted fifth harmonic amplitude, h _inv_5 The inverse calibration function of monolithic stress harmonics is established to correspond to the fifth harmonic.

[0063] Step 205: Compare and verify the compressive stress with the equivalent compressive stress, calculate the deviation between the two. If the deviation exceeds the preset deviation threshold, it is determined that the current extracted data is abnormal, and the excitation current signal is re-acquired. If the deviation does not exceed the preset deviation threshold, it is determined that the current extracted data is valid, and the subsequent steps are continued.

[0064] After obtaining the equivalent compressive stress based on the third harmonic and the verification compressive stress based on the fifth harmonic, the absolute difference or relative ratio between the two is calculated. The preset deviation threshold can be set according to the accuracy requirements of the measurement system, for example, 0.5 MPa. When the deviation is greater than or equal to the preset deviation threshold, it is determined that the harmonic distribution under the current working condition deviates from the magnetoelastic response law of the material, the blocking action is executed, and constant voltage excitation and waveform acquisition are retried.

[0065] To calculate the corresponding hysteresis loss component and air gap additional loss component, this embodiment also proposes a possible implementation method, including:

[0066] Substitute the equivalent compressive stress into the preset stress-corrected hysteresis loss model to calculate the hysteresis loss component under the current state.

[0067] In this embodiment, the hysteresis loss of the core material is affected by the internal magnetic domain state. With increasing clamping force, compressive stress is generated inside the oriented silicon steel sheet, leading to the nucleation and expansion of additional magnetic domains, thereby reducing the magnetoelastic energy. This stress-induced change in the magnetic domain structure increases energy loss during alternating magnetization. To quantify this loss, this embodiment introduces a stress variable as an adjustment parameter for calculation. The calculation logic of the preset stress-corrected hysteresis loss model is as follows:

[0068] P _hyst= k _h (σ)×f×B _m α(σ) ;

[0069] Among them, P _hyst为 Hysteresis loss component, σ is equivalent compressive stress, k _h (σ) is the hysteresis loss coefficient dependent on the equivalent compressive stress, f is the excitation frequency, and B _m α is the peak value of the magnetic flux density of the iron core, and α(σ) is a specific exponential parameter that depends on the equivalent compressive stress.

[0070] Accordingly, k _h σ and α(σ) are not constants, but variables that change dynamically with the stress state. Their sequence values ​​are derived from offline measurements of silicon steel samples from the same batch. Specifically, a controlled compressive stress sequence is applied to the sample on a single-piece magnetic testing instrument. Constant-pressure sinusoidal excitation is performed at each stress level, and the hysteresis loop is measured. The corresponding hysteresis loss value is extracted from the hysteresis loop area, and k is determined by nonlinear least-squares fitting. _h The values ​​of α(σ) and α(σ) at various stress levels. That is, by substituting the equivalent compressive stress into the stress-corrected hysteresis loss model, the hysteresis loss assessment value generated by stress is separated.

[0071] The equivalent air gap length is obtained by converting the air gap characterization index. Using the pre-configured air gap loss coefficient in combination with the equivalent air gap length, the additional air gap loss component that is directly proportional to the air gap length is calculated.

[0072] The air gap characterization index corresponds to the additional fundamental excitation current required to drive the magnetic flux through the air gaps of all lamination seams. After separating this index, it is converted into the equivalent air gap length in physical dimensions. The conversion logic formula is as follows:

[0073] g=ξ _g ×μ _0 ×N / B _m ;

[0074] Where g is the equivalent air gap length, ξ _g μ is a characterization index for air gaps. _0 Where is the vacuum permeability, N is the number of turns in the test winding, and B is... _mThis represents the peak value of the magnetic flux density in the iron core.

[0075] In stacked iron cores, additional air gap losses include the reduction in effective cross-section at the lamination joints and localized eddy current losses caused by edge flux diffusion. Under rated excitation conditions, the increase in this eddy current loss is linearly proportional to the equivalent air gap length, i.e.:

[0076] P _gap= k _g ×g×f 2 ×B 2 _m ;

[0077] Among them, P _gap Add a loss component to the air gap, k _g Here, g is the pre-configured air gap loss coefficient, g is the equivalent air gap length, f is the excitation frequency, and B is the excitation frequency. _m The peak value of the magnetic flux density of the iron core is represented by 2, where 2 in the formula represents the square.

[0078] Based on this, we can obtain hysteresis loss, which independently reflects the destructive effect of clamping stress, and additional air gap loss, which reflects the degree of joint compression.

[0079] Since this coefficient depends on prior physical parameters such as the number of lap stages, lap spacing, lamination thickness, and silicon steel resistivity of the multi-stage step lap structure, in some alternative implementations, a variety of alternative schemes can be used to determine the pre-configured air gap loss coefficient.

[0080] As a first optional implementation example, a corresponding matrix containing core manufacturing design parameters is established. Based on the current core's preset overlapping structure parameters, the matching air gap loss coefficient is retrieved from the corresponding matrix, i.e., the air gap loss coefficient is extracted based on historical statistical experience data.

[0081] As a second alternative implementation example, using a test sample with the same overlap structure as the current core, non-magnetic standard shims of known thickness are inserted at the lamination joints to change the physical length of the air gap. Under the same excitation conditions, the additional losses corresponding to different air gap lengths are measured, and linear regression calculations are performed to extract the slope parameter of the regression line, thereby determining the pre-configured air gap loss coefficient.

[0082] As a third alternative implementation example, a multi-dimensional finite element electromagnetic model of the core joint region is constructed, different air gap length parameters are applied, the corresponding local eddy current loss increment is calculated, and the proportional mapping factor of the calculation result is extracted, i.e., the pre-configured air gap loss coefficient.

[0083] In one possible implementation, for a monotonically increasing pressure application path that gradually increases the clamping force of the stacked cores, acquisition and extraction, decoupling and calculation are performed at each pressure step. Based on the hysteresis loss component and the air gap additional loss component at each pressure step, the total loss gradient is determined, specifically including:

[0084] Step 301: Set a lower limit for the preset pressure step increment so that the increment of the stress characterization index between adjacent pressure steps is higher than the preset system noise threshold.

[0085] Specifically, the stress characterization index is the third harmonic amplitude, and its change with increasing clamping force must be greater than the background electrical noise of the measuring hardware. The preset system noise threshold is set to three times the background noise of the measuring equipment. Based on the pre-calibrated stress response slope, the minimum allowable increase in clamping force is calculated and configured as the lower limit of the preset pressure step increment, so that the calculated loss gradient reflects the actual physical evolution.

[0086] To ensure interpolation accuracy, a preset pressure step constraint is required. The upper limit of this constraint is configured based on engineering experience, and the value is set to ensure that the rated pressure range is scanned within 10 to 30 steps.

[0087] Step 302: Under each preset pressure step, calculate the hysteresis loss gradient based on the hysteresis loss components of the current pressure step and the previous pressure step. That is, unlike the previous method which optimized the total loss through a single-point measurement, this embodiment decomposes the total loss into two gradient components with opposite directions and tracks them separately. Under each preset pressure step, a positively increasing hysteresis loss gradient is calculated based on the hysteresis loss components of the current pressure step and the previous pressure step, and a negative air gap loss gradient with decreasing absolute value is calculated based on the air gap additional loss components of the current pressure step and the previous pressure step.

[0088] Specifically, the gradient change rate is evaluated by using the ratio of the quantization loss difference between adjacent pressure acquisition points to the preset pressure step. Due to the damage effect of compressive stress on the magnetic domains of the iron core material, its hysteresis loss increases, and the corresponding gradient value is always positive and shows an increasing trend.

[0089] Under pressure, the gap between the laminated joints is compacted, the additional loss in the air gap decreases, and the corresponding gradient value is always negative and its absolute value decreases. The corresponding expression is:

[0090] G _hyst_j =(P _hyst_j -P _hyst_j-1 ) / Δ _F ;

[0091] G _gap_j =(P _gap_j -P _gap_j-1 ) / Δ _F ;

[0092] Among them, G _hyst_j P represents the hysteresis loss gradient corresponding to the current pressure step. _hyst_j P represents the hysteresis loss component under the current pressure step. _hyst_j-1 Δ represents the hysteresis loss component under the previous pressure step. _F To preset the pressure step, G _gap_j P represents the air gap loss gradient corresponding to the current pressure step. _gap_j Add a loss component to the air gap at the current pressure step, P _gap_j-1 This is an additional loss component in the air gap from the previous pressure step.

[0093] Step 303: The hysteresis loss gradient and the air gap loss gradient are superimposed to obtain the total loss gradient; wherein the total loss gradient monotonically crosses from negative to positive values.

[0094] The two gradient components in opposite directions are algebraically summed. In the initial low-pressure stage of pressurization, the rate of decrease in marginal loss caused by air gap compression dominates, and the total loss gradient is negative. As the clamping force increases, the marginal increase rate of stress damage to magnetic properties will exceed the marginal benefit brought by air gap closure. The total loss gradient monotonically increases from negative and crosses zero at the preset pressure level, turning into a positive value.

[0095] Furthermore, based on the total loss gradient, the first optimal pressure point on the ascending branch is located, including:

[0096] Step 304: When the total loss gradient is detected to change from negative to positive between two adjacent pressure steps, linear interpolation is performed using the total loss gradients corresponding to the two adjacent pressure steps to locate the first optimal pressure point where the total loss gradient is zero.

[0097] Specifically, the sign of the total loss gradient output at each pressure step is compared in real time. That is, when the total loss gradient of the previous pressure step is less than zero, and the total loss gradient of the current pressure step is greater than or equal to zero, it is determined that the minimum loss range has been captured. The corresponding formula is:

[0098] F _star =F _j-1 +Δ _F ×abs(Gal _j-1 ) / (abs(Gal _j-1 )+Gal _j );

[0099] Among them, F _star F is the first optimal pressure point determined by interpolation. _j-1 Δ is the clamping force value of the previous pressure step. _F For the preset pressure step, abs is the absolute value function, Gal _j-1Gal represents the total loss gradient of the previous pressure step. _j This represents the total loss gradient for the current pressure step.

[0100] Furthermore, to prevent the pinning and solidification of silicon steel magnetic domains from causing delayed branch switching, voltage reduction and callback are prohibited.

[0101] Step 305: Calculate the gradient change rates of the hysteresis loss gradient and the air gap loss gradient in the neighborhood of the first optimal pressure point, and combine them as the local loss change rate of the corresponding first optimal pressure point.

[0102] After locating the extreme point, extract the gradient difference data of the adjacent steps on both sides of the extreme point to evaluate the curvature characteristics of the loss surface at the bottom. The corresponding expression is:

[0103] G _hyst_prime =(G _hyst_j -G _hyst_j-1 ) / Δ _F ;

[0104] G _gap_prime =(G _gap_j -G _gap_j-1 ) / Δ _F ;

[0105] Among them, G _hyst_prime G is the rate of change of the hysteresis loss gradient. _hyst_j G represents the hysteresis loss gradient corresponding to the current pressure step. _hyst_j-1 G represents the hysteresis loss gradient corresponding to the previous pressure step. _gap_prime G is the rate of change of the air gap loss gradient. _gap_j G represents the air gap loss gradient corresponding to the current pressure step. _gap_j-1 Δ represents the air gap loss gradient corresponding to the previous pressure step. _F The preset pressure step is used. Based on this, a set of gradient rate parameters is derived, which constitutes the local loss rate.

[0106] In engineering implementation, to balance search time and interpolation accuracy, in another specific embodiment, a two-stage variable step size strategy is adopted instead of a fixed step size strategy, including:

[0107] First, perform a coarse search with a step increment until the absolute value of the total loss gradient is detected to be lower than a preset neighborhood threshold; the preset neighborhood threshold can be determined according to the gradient measurement resolution of the system and the process cycle requirements.

[0108] For example, in the initial stage, a large first step increment is used to drive the servo mechanism to quickly increase the pressure, and the convergence of the gradient is monitored. A preset neighborhood threshold is set as a positive tolerance extreme value. If the absolute value of the current gradient is less than the preset neighborhood threshold, and the gradient sign is still negative, it is determined that the current pressure has approached the flat region to the left of the optimal value, and the search at this level is paused to prevent crossing the zero point.

[0109] The pressure step is switched to a second step increment that is smaller than the first step increment. A fine search is performed in the pressure interval where the absolute value of the total loss gradient is lower than the neighborhood threshold until a crossover where the total loss gradient turns from negative to positive is detected, which is used to locate the first optimal pressure point.

[0110] Accordingly, upon entering the extreme value sensitive region, the increment of the executed instruction is reduced to the second step increment. The second step increment is set to 1 / 4 or lower of the first step increment. In high-resolution mode, unidirectional boosting continues until a node is reached where the gradient changes from negative to positive.

[0111] In one possible implementation, preset compensation parameters are introduced, including:

[0112] The hysteresis asymmetric parameter characterizing the difference in the pressure increase and decrease branch gradients of the material, and the pressure relaxation parameter characterizing the influence of the curing process, are obtained together as compensation parameters.

[0113] In this embodiment, due to the delayed recovery of the pinned supplementary magnetic domains during voltage reduction in oriented silicon steel, the response rate of hysteresis loss along the voltage reduction direction is lower than that along the voltage increase direction. Therefore, a hysteresis asymmetry parameter is introduced to quantify this asymmetry, and its value can be determined by performing cyclic loading tests including voltage increase and decrease phases on samples from the same batch.

[0114] After the iron core is cured under a preset insulating coating material and curing temperature profile, mechanical stress release occurs. Pressure relaxation parameters are obtained to characterize the amount of clamping force reduction caused by this process.

[0115] Based on the hysteresis loss gradient and hysteresis asymmetric parameters corresponding to the first optimal pressure point, the magnetic coupling asymmetric compensation term is calculated.

[0116] The calculation of the magnetic coupling asymmetry compensation term is used to counteract the extreme point drift caused by magnetic hysteresis asymmetry. This drift manifests as the optimal total loss pressure on the step-down branch deviating from the optimal pressure on the step-up branch due to hysteresis. The formula for calculating the magnetic coupling asymmetry compensation term is:

[0117] C _asym =(1-η _mm )×G _hyst_star ;

[0118] Among them, C _asymFor the magnetic coupling asymmetry compensation term, η _mm G is a hysteresis asymmetric parameter. _hyst_star This represents the hysteresis loss gradient corresponding to the first optimal pressure point.

[0119] Based on the local loss change rate, hysteresis asymmetric parameter, and pressure relaxation parameter, a relaxation offset compensation term characterizing the effect of pressure relaxation and a local curvature evaluation term characterizing the surface features near the extreme point are calculated, respectively.

[0120] Specifically, for the secondary offset caused by relaxation during curing, this embodiment uses the local loss change rate to extract the surface morphology near the extreme point. The relaxation offset compensation term includes contributions to compensate for the hysteresis loss gradient offset on the pressure-reducing branch caused by pressure relaxation, and additional losses caused by the air gap reopening due to relaxation. The local curvature evaluation term is calculated and used as the convergence factor for the entire asymmetric compensation calculation; the corresponding expression is:

[0121] C _relax =η _mm ×G _hyst_prime ×Δ _relax +G _gap_prime ×Δ _relax ;

[0122] C _curve =η _mm ×G _hyst_prime +G _gap_prime ;

[0123] Among them, C _relax For the relaxation offset compensation term, η _mm G is a hysteresis asymmetric parameter. _hyst_prime Let Δ be the local loss rate of the hysteresis loss gradient in the neighborhood of the first optimal pressure point. _relax G is the pressure relaxation parameter. _gap_prime C represents the local loss rate of the air gap loss gradient within the neighborhood of the first optimal pressure point. _curve This is a local curvature evaluation term.

[0124] The offset compensation amount is determined based on the superposition value of the magnetic coupling asymmetry compensation term and the relaxation offset compensation term, combined with the local curvature evaluation term, including:

[0125] The absolute value of the local curvature evaluation term is determined to be lower than a preset curvature threshold. When the local curvature evaluation term approaches 0, it indicates that the pressure-reducing branch of the loss surface near the first optimal pressure point is nearly curvature-free, i.e., flat. The absolute value of the local curvature evaluation term is extracted and compared with a preset tolerance benchmark, i.e., the preset curvature threshold. The preset curvature threshold can be determined through numerical stability analysis based on the order of magnitude of the measurement system's resolution and the expected offset compensation.

[0126] If the curvature is below the curvature threshold, the current linearization compensation model is deemed to have failed, and an abnormal reassessment action is triggered. The abnormal reassessment action includes re-executing gradient tracking and positioning with a reduced pressure step and redetermining the offset compensation amount.

[0127] When the absolute value is less than the curvature threshold, the algebraic equation is deemed to have encountered singular boundary conditions. Specifically, to stabilize the industrial control system, it is necessary to block conventional division calculations and generate a status signal to trigger an anomaly reassessment. The anomaly reassessment can be to pause the automated calculation process, switch to a small-step fine-tuning mode, or measure the physical loss data of the operating point after relaxation on the buck branch.

[0128] If the value is not lower than the curvature threshold, the sum of the magnetic coupling asymmetry compensation term and the relaxation offset compensation term is divided by the local curvature evaluation term to obtain the offset compensation amount. Specifically, when the absolute value is not lower than the curvature threshold, a complete superposition division operation is performed, i.e.:

[0129] Δ _F_corr =(C _asym +C _relax ) / C _curve ;

[0130] Where, Δ _F_corr C is the offset compensation amount. _asym For the magnetic coupling asymmetry compensation term, C _relax For the relaxation offset compensation term, C _curve This is a local curvature evaluation term.

[0131] Step 104e: The offset compensation amount is superimposed on the first optimal pressure point to obtain the target curing pressure.

[0132] After obtaining the offset compensation amount, perform terminal numerical synthesis calculation, using the following formula:

[0133] F _opt =F _star +Δ _F_corr ;

[0134] Among them, F _opt For the target curing pressure, F _star For the first optimal pressure point, Δ _F_corr This is the offset compensation amount. After offset mapping and coordinate translation, the target curing pressure is generated, which can take into account the irreversible hysteresis characteristics of the material and the stress release characteristics of the process.

[0135] In one possible implementation, locking the stacked iron cores according to the target curing pressure also involves overshoot identification and closed-loop control, specifically including:

[0136] Step 401: Obtain the clamping force currently reached under the monotonically increasing pressure application path; determine whether the target curing pressure is less than the currently reached clamping force; if it is not less than the currently reached clamping force, control the clamping mechanism to continue increasing the pressure to the target curing pressure and perform locking; if it is less than the currently reached clamping force, determine that an overshoot state has occurred that violates the monotonically increasing constraint, stop increasing the pressure, and use the currently reached clamping force as the upper limit of the actual curing pressure to perform locking evaluation; based on the locking evaluation result, determine whether to use the currently reached clamping force as the curing pressure to perform locking.

[0137] Specifically, during the search for the optimal pressure on the ascending branch, the actuator increases the pressure step by step. When a gradient crossing is detected at step j, the target curing pressure F is calculated. _opt If the calculated F is affected by a large step size or a sudden change in local curvature, _opt Less than the clamping force F currently applied to the iron core _j Therefore, if the actuator wants to achieve F... _opt A pressure reduction measure is required.

[0138] Because the magnetic domain response of grain-oriented silicon steel exhibits historical dependence, voltage reduction operations can easily cause the core state to enter unknown loss branches, rendering the established decoupling model ineffective. The system extracts the currently reached clamping force F in real time. _j and with the target curing pressure F _opt Perform numerical comparison. If F is satisfied... _opt Not less than F _j If the condition is met, the actuator will continue pressurizing after the target value is at the current position. If this condition is not met, the system is determined to have experienced a search overshoot, and the pressure increase will be immediately stopped, initiating a safety check for the overshoot condition.

[0139] This embodiment is used to eliminate the conflict between the target solidification pressure and the monotonicity constraints during the execution process.

[0140] Step 401a: Calculate the estimated total loss at the working point of the pressure reduction branch after the stacked iron core is locked with the currently reached clamping force and the pressure is released according to the pre-configured pressure relaxation parameters.

[0141] Step 401b: Based on the target curing pressure and the pre-configured pressure relaxation parameters, the theoretical minimum loss is determined using the same calculation method as the hysteresis loss component and the air gap additional loss component.

[0142] Step 401c: Compare the estimated total loss with the theoretical minimum loss determined based on the target curing pressure;

[0143] Step 401d: If the deviation between the estimated total loss and the theoretical minimum loss is within the preset acceptable range, the preset acceptable range can be determined according to the no-load loss design margin requirements of the transformer. The clamping force that has been reached is determined as the final solidification pressure and locking is performed.

[0144] If the deviation exceeds the preset acceptable range, an overshoot alarm signal will be output, and the preset pressure step will be reduced before re-executing the monotonically increasing pressure application path from the initial preload.

[0145] In this embodiment, the loss deviation caused by overshoot is quantified by locking the evaluation logic. Due to the overshoot point F... _j This is already the peak stress point on the control path; that is, regardless of the curing relaxation process, the effective working pressure of the core will decrease from F. _j Starting from there, it slides down the step-down branch. This embodiment utilizes the calibrated characteristics of the step-down branch to calculate the loss value after locking at the current overshoot point and then relaxing.

[0146] Among them, G _total_j_down The initial total loss gradient at the current overshoot point on the buck branch is calculated based on the measured gradient of the boost branch and the hysteresis asymmetric parameters, specifically:

[0147] G _total_j_down =η _mm ×G _hyst_j +G _gap_j ;

[0148] In this formula, G _hyst_j With G _gap_j These are the measured hysteresis loss gradient and air gap loss gradient of the current pressure step on the boost branch, respectively.

[0149] P _real =P _total_j +(η _mm ×G _hyst_prime +G _gap_prime )×(Δ _relax 2 / 2)-Gal _j_down ×Δ _relax ;

[0150] Among them, P _real P is the estimated total loss at the operating point of the step-down branch. _total_j η represents the measured total loss corresponding to the current clamping force. _mm G is a hysteresis asymmetric parameter. _hyst_prime G represents the local rate of change of the hysteresis loss gradient. _gap_prime Δ represents the local rate of change of the air gap loss gradient. _relax For pressure relaxation parameters, Gal _j_downThis represents the initial total loss gradient at the current point on the buck branch, with the superscript 2 indicating that it is squared.

[0151] Gal _j_down Based on the hysteresis response characteristics of the pressure-drop branch, the hysteresis loss gradient of the current pressure step is determined by the hysteresis asymmetric parameter η. _mm The sum of the scaled value and the air gap loss gradient of the current pressure step.

[0152] Target curing pressure F _opt The theoretical minimum loss P obtained after locking and relaxation _min Estimated total loss P _real With P _min Subtract the values ​​to obtain the deviation value. If this deviation value is less than the preset acceptable range, for example, less than 0.5%, adjust the current clamping force F. _j Confirm the curing pressure and execute the lock.

[0153] In one optional implementation, if the deviation exceeds a preset acceptable range, it indicates a large overshoot, leading to a deterioration in loss optimization. Based on this, the system outputs an alarm signal and performs adaptive adjustment. The adaptive adjustment can be to reduce the preset pressure step Δ... _F Value, reinitialize the actuator, and start from the initial preload F _0 A new round of monotonic boost search is initiated to relocate the optimal value with a finer sampling resolution.

[0154] Furthermore, before acquiring the excitation current signal of the stacked iron core under constant voltage sinusoidal excitation conditions, the offline construction of pre-configured parameters is also included, specifically:

[0155] Step 100a: Using the same batch of samples as the stacked iron core, perform constant pressure excitation test under gapless conditions. During the test, apply controlled compressive stress, extract the mapping relationship between stress level and harmonic characteristics and fundamental wave characteristics, and construct pre-configured single-piece stress harmonic calibration function and pre-store gapless magnetization fundamental wave component respectively.

[0156] Specifically, the sample preparation used grain-oriented silicon steel coils of the same material, grade, and batch as the stacked iron core, cut into standard samples according to the rolling direction. The testing equipment employed a single-piece magnetic tester equipped with a stress-loading fixture. During testing, the fixture applied a uniformly distributed in-plane compressive stress to the sample, and constant-voltage sinusoidal excitation was performed using an AC adjustable power supply.

[0157] Accordingly, the stress loading process increases in preset equal-interval steps. At each stress level, after the stress stabilizes, the excitation current is collected, and the third harmonic amplitude is extracted. Multiple stress levels are curve-fitted with their corresponding third harmonic amplitudes to construct a single-piece stress harmonic calibration function, the expression of which is:

[0158] h(σ) = a × (exp(b × σ) - 1) + c;

[0159] Where h(σ) is the third harmonic amplitude obtained by fitting, σ is the applied compressive stress, a is the proportionality coefficient, exp is the exponential function with the natural constant as the base, b is the growth rate parameter, and c is the initial harmonic offset under zero stress.

[0160] During the same test, the fundamental current component corresponding to each stress level is recorded and used as the pre-stored air-gap-free magnetized fundamental current component. This component is stored in the control system's storage module in the form of a data table for real-time retrieval based on the equivalent compressive stress during online operation.

[0161] Step 100b: Perform a cyclic loading test on the sample, including boost and buck phases. By comparing the loss response differences between the buck and boost branches, extract the pre-configured hysteresis asymmetric parameters.

[0162] The sample was subjected to a complete stress loading cycle on a single-piece magnetic testing instrument. Specifically, the compressive stress was slowly increased from zero to a preset peak level, and the hysteresis loss gradient during this pressure increase process was recorded; then, the pressure was slowly decreased from the peak level, and the hysteresis loss gradient during this pressure decrease process was recorded. The hysteresis asymmetry parameter is defined as the ratio of the pressure decrease branch gradient to the pressure increase branch gradient within the neighborhood of the target stress, specifically:

[0163] η _mm =G _down / G _up ;

[0164] Where, η _mm G is a hysteresis asymmetric parameter. _down G represents the hysteresis loss gradient of the voltage reduction branch at the target stress point. _up This represents the hysteresis loss gradient of the boost branch at the same stress point.

[0165] Step 100c: Obtain the physical overlap parameters at the joint of the stacked iron core laminations, and extract the pre-configured air gap loss coefficient that matches the physical overlap parameters based on the experience database.

[0166] In this embodiment, the physical overlap parameters include the number of overlap levels, the step spacing, and the stack thickness. For multi-level step overlap structures, the additional air gap loss depends on the electromagnetic field distribution at the joint. The number of overlap levels is determined according to the design drawings, for example, 5 overlap levels, and the corresponding coefficients are retrieved from a pre-built empirical database.

[0167] In the absence of historical data, the three-dimensional finite element numerical simulation method is used to calculate the eddy current loss increment under different air gap lengths, and the pre-configured air gap loss coefficient is obtained by reverse calculation.

[0168] Step 100d: Obtain the characteristic parameters of the insulation coating of the stacked iron core and the target curing process state, and determine the pre-configured pressure relaxation parameters before and after curing based on the calculation model.

[0169] The pressure relaxation parameter reflects the pressure drop of the core from the locked state to the completion of curing. In this embodiment, it is calculated based on the relaxation modulus of the insulating coating and the curing temperature curve, using the following formula:

[0170] Δ _relax =(F _initial ×S _factor )×(1-exp(-t _process / τ _c ));

[0171] Where, Δ _relax F is the pressure relaxation parameter. _initial S is the pressure applied during locking. _factor t is a geometric factor related to the coating contact area. _process τ is the duration of the curing process. _c This is the characteristic relaxation time constant of the material. In engineering, this parameter can be obtained by statistically comparing the pressure of the iron core before and after curing under the same process conditions.

[0172] With the pre-construction of offline parameters, the online control algorithm can obtain complete data support. During offset compensation, the superposition of the magnetic coupling asymmetric compensation term and the relaxation offset compensation term makes the target curing pressure higher than the first optimal pressure point, and allows the effective working pressure of the core after curing and relaxation to fall back to the neighborhood of the minimum loss value of the pressure reduction branch.

[0173] Furthermore, the constant voltage sinusoidal excitation condition is achieved by an AC power supply with low output impedance; the short-circuit impedance of the AC power supply is less than 1% of the excitation impedance of the stacked iron core, so that the external voltage distortion caused by the excitation current harmonics is lower than the preset voltage distortion threshold.

[0174] Based on the same inventive concept as the above-described method embodiments, this embodiment also provides an adaptive control system for reducing transformer no-load losses by stacking core pressure. This control system can be used to execute the control method provided in the above embodiments.

[0175] In this embodiment, the control system includes a hardware execution subsystem and a logic control subsystem. The hardware execution subsystem includes a servo electric actuator, a low-output-impedance AC power supply, and a high-frequency data acquisition device. The servo electric actuator is mounted at the core stacking station and applies a continuously adjustable, monotonically increasing clamping force to the stacked cores. It is equipped with a high-precision pressure sensor to provide real-time feedback on the current physical clamping force. The low-output-impedance AC power supply provides a constant-voltage sinusoidal excitation voltage to the test winding of the stacked cores. To prevent distortion of the applied voltage waveform that could affect the test, the short-circuit impedance of the low-output-impedance AC power supply is limited to less than 1% of the excitation impedance of the stacked cores. The high-frequency data acquisition device is connected to the test winding and synchronously acquires the real-time excitation current signal at a sampling rate 20 times higher than the fundamental frequency.

[0176] The logic control subsystem is implemented using a programmable logic controller or an industrial control computer. It is equipped with multiple functional modules, including a data acquisition and decoupling module, a loss quantification and tracking module, and a compensation calculation and execution module.

[0177] The data acquisition and decoupling module is connected to a high-frequency data acquisition device to acquire the excitation current signal of the stacked iron core under constant voltage sinusoidal excitation conditions, and uses the discrete Fourier algorithm to extract the harmonic characteristic components of the excitation current signal.

[0178] Furthermore, this module utilizes the air gap invariance of higher harmonics to decouple the harmonic characteristic components into stress characterization indicators and air gap characterization indicators.

[0179] The loss quantification and tracking module receives stress characterization indicators and air gap characterization indicators, calls a preset physical mapping model, and calculates the corresponding hysteresis loss components and air gap additional loss components. During the dynamic process of the servo electric actuator monotonically increasing the applied pressure, this module extracts the total loss gradient based on the hysteresis loss components and air gap additional loss components at each pressure step, and locates the first optimal pressure point on the rising branch based on the zero-point crossing position characteristics of the total loss gradient.

[0180] The compensation calculation and execution module is used to obtain pre-configured compensation parameters, perform offset compensation calculations on the first optimal pressure point based on the compensation parameters, and output the target curing pressure. After performing monotonicity constraint comparison and confirming that no search overshoot has occurred, this module sends a closed-loop control command to the servo electric actuator to mechanically lock the stacked iron core according to the target curing pressure.

[0181] This embodiment also provides a core stacking pressure control system for reducing transformer no-load losses. The system includes a processor coupled to a memory. The memory stores instructions. When the instructions are executed by the processor, they can implement all the steps of any of the control methods in the above embodiments.

[0182] Based on the same inventive concept as the above-described method embodiments, this embodiment also provides a computer-readable storage medium storing a computer program thereon. When the computer program is invoked and executed by a control unit or processor, all steps of any of the control methods described above can be implemented.

[0183] Through the coordinated operation of hardware and software modules, the core stacking pressure adaptive control system provided in this embodiment can be freed from dependence on the embedded physical stress sensor inside the core. It can achieve adaptive optimization and asymmetric hysteresis compensation of mechanical pressure by utilizing the electrical harmonic characteristics of the test terminals, thereby realizing the global minimization control of transformer no-load loss in batches and automatically in industrial production sites.

[0184] According to another aspect of this application, this embodiment also provides an adaptive control method for core stacking pressure to reduce transformer no-load losses, comprising:

[0185] An initial preload is applied to the stacked iron core, and the excitation current signal under constant pressure sinusoidal excitation is acquired. The harmonic characteristic components of the excitation current signal are extracted, and the harmonic characteristic components are decoupled into stress characterization index and air gap characterization index by utilizing the air gap invariance of higher harmonics. Based on the stress characterization index and air gap characterization index, the hysteresis loss component and air gap additional loss component under the current state are calculated. Under the monotonically increasing pressure application path, the clamping force is gradually increased in a preset pressure step. The corresponding hysteresis loss component and air gap additional loss component are acquired at each pressure step. The total loss gradient formed by the superposition of the increase in hysteresis loss component and the decrease in air gap additional loss component of adjacent pressure steps is tracked. The zero-point crossing position of the total loss gradient is detected to locate the first optimal pressure point on the rising branch. The pre-configured hysteresis asymmetric parameter and the pre-configured pressure relaxation parameter are acquired. The first optimal pressure point is compensated based on the hysteresis asymmetric parameter and the pressure relaxation parameter to obtain the target curing pressure. The stacked iron core is locked according to the target curing pressure.

[0186] As an alternative implementation of this application, this embodiment provides a second adaptive control method for core stacking pressure to reduce transformer no-load losses, including:

[0187] The stacked transformer cores are installed in the stacking station of the adjustable clamping mechanism, and an initial preload F is applied. _0 The force sensor confirmed F _0 After establishment, record the effective cross-sectional area A of the iron core. _c Effective magnetic circuit length l _c and the number of turns N in the winding.

[0188] A sinusoidal excitation voltage with a frequency of power frequency f and an amplitude of the rated value is applied to the iron core test winding through a low-output-impedance AC voltage regulator. Under this condition, the peak value of the iron core magnetic flux density B...m Determined by the applied voltage, i.e.

[0189] B _m =V _m / (ω×N×A _c );

[0190] Where ω = 2πf is the angular frequency, V m The voltage amplitude is denoted as t. In this embodiment, the excitation current waveform i(t) flowing through the winding is synchronously acquired at a sampling rate of not less than 20 times the fundamental frequency, with each acquisition covering no less than 10 complete power frequency cycles.

[0191] Accordingly, a discrete Fourier transform is performed on i(t) to extract the fundamental amplitude and the amplitudes of the third and fifth harmonics, forming a reference harmonic eigenvector, which serves as the starting reference for decoupling and gradient tracking.

[0192] Furthermore, under sinusoidal voltage excitation conditions, the magnetic flux density inside the iron core is:

[0193] B(t) = B _m sin(ωt);

[0194] The excitation current satisfies Ampere's circuital law, or in other words, the physical model of the decoupling logic, and the corresponding formula is:

[0195] N×i(t)=H(B(t),σ)×l _c +(B(t)×g) / μ _0 ;

[0196] Where H(B,σ) represents the magnetic field strength-magnetic flux density relationship of the grain-oriented silicon steel under σ, g is the total equivalent air gap length at all lamination joints of the iron core, and μ _0 Let be the vacuum permeability. The first term on the right-hand side contains harmonic components due to the nonlinearity of the HB relationship; the second term on the right-hand side expands to (B... _m ×g / μ _0 sin(ωt) is a linear function of magnetic flux density and is a fundamental sinusoidal signal. Its Fourier expansion contains only the fundamental component and does not produce higher harmonics.

[0197] After performing Fourier decomposition on the excitation current, the amplitude of the nth odd harmonic (n≥3) is determined by the nonlinear magnetization characteristics of the core material, namely:

[0198] I _n =l _c / N×∣H _n (B _m ,σ)∣;

[0199] Among them, H _n (B _m ,σ) is H(B_m The nth Fourier coefficient of sin(ωt),σ, excluding g, means that under constant voltage excitation conditions, the amplitude of higher harmonics of the excitation current is invariant to the size of the air gap in the lamination. Therefore, the amplitude of the third harmonic is set as a purely stress-sensitive index, i.e.:

[0200] ξ _σ =I _3 ;

[0201] In oriented silicon steel, as compressive stress increases, stress-induced supplementary magnetic domains cause increased deformation at the knee of the magnetization curve, leading to a monotonically increasing third harmonic amplitude. The fundamental component's contribution to the material's magnetization and the air gap is specifically as follows:

[0202] I _1 =∣l _c / N⋅H _1 (B _m ,σ)+(B _m ×g) / (μ _0 ×N)∣;

[0203] Using ξ _σ The air gap invariance is used to quantify and estimate the equivalent compressive stress currently borne by the iron core, and to separate the air gap contribution in the fundamental frequency. The monolithic stress-harmonic calibration function h(σ) matching the current silicon steel grade is read. On a monolithic magnetic testing instrument, controlled compressive stress is applied to silicon steel samples from the same batch, under air gap-free conditions and with the same B... _m The horizontal measurement corresponds to the third harmonic amplitude, obtained through testing and calibration. Regarding ξ... _σ Perform the inverse function operation to obtain the estimated equivalent compressive stress σ. * :

[0204] σ * =h -1 (ξ _σ );

[0205] Accordingly, the stress σ is retrieved from the calibration data. * At that time, the fundamental magnetization current component I under the condition of no air gap _1,core(σ*) Using the measured fundamental frequency I _1 Subtracting this component, the air gap sensitivity index ξ is separated. _g ξ _g This corresponds to the additional fundamental excitation current required to drive the magnetic flux through the entire air gap of the laminations. Based on this, the estimated equivalent air gap length g is calculated. * The calculation formula is:

[0206] g * =(ξ _g ×μ _0 ×N) / B _m ;

[0207] The decoupled σ * and g * Substitute each component into the corresponding loss component model for quantification. Stress-related hysteresis loss P _hyst Characterized by the Steinmetz equation, the corresponding expression is:

[0208] P _hyst (σ * )=k _h (σ * )×f×B _m α(σ*) ;

[0209] Where, k _h (σ * α and α(σ*) are the hysteresis loss coefficient and the Steinmetz exponent, respectively, derived from single-chip calibration experiments. Air gap additional loss P _gap The local eddy current loss, caused by the reduction in effective cross-section at the lamination joint and the diffusion of magnetic flux at the edges, is approximately proportional to the equivalent air gap length.

[0210] P _gap (g * )=k _g ×g * ×f 2 ×B _m 2 ;

[0211] Where, k _g The air gap loss coefficient related to the core step-over joint structure parameters is used to obtain the decoupling loss assessment result under the current clamping force [P]. _hyst ,P _gap ].

[0212] Accordingly, a monotonically increasing constraint is established for the pressure search. When grain-oriented silicon steel is subjected to compressive stress, supplementary magnetic domains nucleate and expand within the grains to reduce magnetoelastic energy; when the stress is released, some of the supplementary magnetic domains are pinned by grain boundary defects and inclusions, making it difficult for them to revert to their initial state. If an alternating ascending and descending search strategy is adopted, such as the bisection method, each pressure reversal causes the magnetic domain state to enter a different magnetic coupling hysteresis branch.

[0213] Set the pressure search to start from F. _0 Starting with a monotonically increasing sequence, set the pressure step increment Δ _F Its value should be such that the change in the third harmonic between adjacent steps is not less than three times the background noise level of the measurement system. During the search process, the clamping force only increases and does not decrease, so that all measurement data are located on the same rising branch of the magnetic coupling hysteresis curve.

[0214] Starting from step j=1, perform progressively increasing pressure acquisition. In each step, increase the clamping force by Δ. _F ,Right now:

[0215] j =F _0 +j×Δ _F ;

[0216] After the pressure sensor reading stabilizes, or after a waiting time of at least three times the stress homogenization time constant between the laminations, repeat the constant pressure excitation current acquisition and Fourier harmonic extraction to obtain the harmonic feature vector of step j.

[0217] H (j) =[I _1 (j) ,I _3 (j) ),I _5 (j) ];

[0218] For H (j) Perform the decoupling and loss quantization process to obtain the decoupling loss [P] at step j. _hyst (j) ,P _gap (j) ].

[0219] Using the decoupling loss data from two adjacent steps, the hysteresis loss gradient and air gap loss gradient are calculated respectively:

[0220] G _hyst (j) =(P _hyst (j) -P _hyst (j-1) / Δ _F ;

[0221] G _gap (j) =(P _gap (j) -P _gap (j-1) ) / Δ _F ;

[0222] Among them, G _hyst (j) It is a positive value and gradually increases with increasing pressure, reflecting the accelerating marginal trend of compressive stress damaging magnetic properties; G _gap (j) The value is negative and its absolute value gradually decreases with increasing pressure, reflecting the diminishing marginal loss benefit from further compression of the air gap. Adding the two values ​​together yields the total loss gradient:

[0223] G _total(j)= G _hyst (j) +G _gap (j) ;

[0224] In the initial low-pressure phase, the rate of decrease in air gap compression loss is dominant, G _total (j) <0; As pressure continues to increase, the rate of increase in stress damage exceeds the rate of increase in air gap gain, G _total (j) It monotonically increases from a negative value and exceeds zero. When two adjacent steps satisfy G... _total (j-1) <0 and G _total (j-1) When the value is ≥0, the total loss reaches its minimum value and crosses zero. Linear interpolation is then performed on the gradient zero point to locate the optimal pressure F of the ascending branch. * The calculation formula is:

[0225] F * =F (j-1) +Δ _F ×∣G _total (j-1) ∣ / (∣G _total (j-1) |+G _total (j) );

[0226] Terminate boost scan and save F * And G in the extreme neighborhood _hyst * G _gap * I _3 * By utilizing the gradient difference between adjacent steps on both sides of the extreme value, the loss surface at F is calculated. * The gradient rate of change at point is specifically:

[0227] G _hyst ' =(G _hyst (j) -G _hyst (j-1) ) / Δ _F ;

[0228] G _gap ' =(G _gap (j) -G _gap (j-1) ) / Δ _F ;

[0229] Introducing the hysteresis asymmetry parameter η to describe the degree of hysteresis asymmetry in magnetic coupling._mm That is, in F * Near the corresponding stress level, the ratio of the hysteresis loss gradient in the step-down branch to that in the step-up branch is typically between 0.4 and 0.8. This parameter can be pre-calibrated through step-up and step-down cyclic experiments on a single sample. η _mm The value <1 reflects the delayed recovery of the pinned supplementary magnetic domains during voltage reduction, resulting in a lower response rate of hysteresis loss along the voltage reduction direction compared to the voltage increase direction. The expected pressure relaxation δ of the core under curing process conditions is read from the process parameter database. _relax This value is determined by the viscoelastic parameters of the insulating coating and the curing temperature curve.

[0230] Under a certain pressure, after the iron core has solidified, the effective clamping force during actual operation will relax δ downwards from the solidification pressure along the pressure-reducing branch. _relax Since the hysteresis loss gradient on the buck branch is equal to η on the boost branch... _mm The working point after relaxation shifts relative to the optimal point of the rising branch.

[0231] To ensure that the actual working pressure after relaxation falls at the minimum position of total loss on the pressure-reducing branch, the curing pressure is changed from F. * Offset upwards by a correction amount ΔF _corr Let the operating point F on the step-down branch be... _opt -δ _relax The total loss gradient is zero, using G _hyst * G _gap * G _hyst ' G _gap ' and η _mm and δ _relax Combining the gradient equilibrium equation of the pressure reduction branch and solving it, the calculation formula is as follows:

[0232] ΔF _corr =(1-η _mm )×G _hyst * +G _gap ' ×δ _relax ) / (η _mm ×G _hyst ' +G _gap ' );

[0233] Among them, (1-η _mm )×G _hyst *Compensation for the asymmetry of magnetic coupling hysteresis. That is, no pressure relaxation occurs, and the optimal pressure of the pressure-reducing branch is higher than the optimal pressure of the pressure-boosting branch due to the hysteresis effect; G _gap ' ×δ _relax The additional losses caused by the air gap reopening due to compensating for relaxation are calculated. The denominator reflects the loss surface at F. * The local curvature characteristics in the vicinity determine the degree of amplification or convergence of the correction amount.

[0234] The corrected optimal curing clamping force F _opt The output is sent to the clamping mechanism control system, specifically as follows:

[0235] _opt =F * +ΔF _corr ;

[0236] The actuator adjusts the core clamping force to F. _opt Then it is locked and enters the curing process. After curing is completed and stress relaxation and stabilization are achieved, the actual effective pressure of the iron core converges to F. _opt -δ _relax Nearby, the operating point is located in the neighborhood of the minimum total loss on the step-down branch, which means that the global optimization of the no-load loss of the iron core is achieved under the condition of considering the irreversible magnetic coupling characteristics of the material and the process relaxation effect.

[0237] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A core stacking pressure control method for reducing no-load loss of a transformer, characterized by, include: Obtain the excitation current signal of the stacked iron core under constant voltage sinusoidal excitation conditions; Harmonic characteristic components of the excitation current signal are extracted, and the harmonic characteristic components are decoupled into stress characterization index and air gap characterization index. The corresponding hysteresis loss component and air gap additional loss component are calculated. Under the monotonically increasing pressure application path, the clamping force of the stacked iron core is gradually increased. Acquisition and extraction, decoupling and calculation are performed in each pressure step. Based on the hysteresis loss component and the air gap additional loss component in each pressure step, the total loss gradient is determined, and the first optimal pressure point on the rising branch is located according to the total loss gradient. Obtain the compensation parameters, and perform offset compensation calculation on the first optimal pressure point based on the compensation parameters to obtain the target solidification pressure; The stacked iron cores are locked according to the target curing pressure.

2. The method according to claim 1, characterized in that, Harmonic characteristic components of the excitation current signal are extracted, including: Discrete Fourier decomposition is performed on the excitation current signal to extract the fundamental amplitude and at least one odd-order higher harmonic amplitude. The fundamental frequency amplitude and the amplitudes of odd-order higher harmonics are used together as harmonic characteristic components.

3. The method according to claim 2, characterized in that, The harmonic characteristic components are decoupled into stress characterization indicators and air gap characterization indicators, including: The third harmonic amplitude among the odd-order higher harmonic amplitudes is used as a stress characterization index. Based on the single-piece stress harmonic calibration function, the third harmonic amplitude is mapped to the equivalent compressive stress. Based on the equivalent compressive stress, obtain the corresponding pre-stored gapless magnetization fundamental wave component; The air gap characterization index is obtained by subtracting the air gap-free magnetized fundamental wave component from the fundamental wave amplitude.

4. The method according to claim 3, characterized in that, It also includes cross-consistency checks, specifically: The fifth harmonic amplitude among the odd-order higher harmonic amplitudes is mapped to the verification compressive stress through the corresponding pre-configured single-piece stress harmonic calibration function. The compressive stress is compared with the equivalent compressive stress, and the deviation between the two is calculated. If the deviation exceeds the preset deviation threshold, it is determined that the current extracted data is abnormal, and the excitation current signal is re-acquired. If the deviation does not exceed the preset deviation threshold, it is determined that the current extracted data is valid, and the subsequent steps are continued.

5. The method according to claim 3, characterized in that, Calculate the corresponding hysteresis loss component and air gap additional loss component, including: Substitute the equivalent compressive stress into the preset stress-corrected hysteresis loss model to calculate the hysteresis loss component under the current state. The equivalent air gap length is obtained by converting the air gap characterization index. Using the pre-configured air gap loss coefficient in combination with the equivalent air gap length, the additional air gap loss component that is directly proportional to the air gap length is calculated.

6. The method according to claim 1, characterized in that, Based on the hysteresis loss component and the air gap additional loss component at each pressure step, the total loss gradient is determined, including: Set a lower limit for the preset pressure step increment so that the increment of the stress characterization index between adjacent pressure steps is higher than the preset system noise threshold. At each preset pressure step, the hysteresis loss gradient is calculated based on the hysteresis loss components of the current pressure step and the previous pressure step. The air gap loss gradient is calculated based on the air gap additional loss between the current pressure step and the previous pressure step. The total loss gradient is obtained by superimposing the hysteresis loss gradient and the air gap loss gradient.

7. The method according to claim 1, characterized in that, Locking the stacked iron cores according to the target curing pressure includes: Obtain the clamping force currently reached under a monotonically increasing pressure application path; Determine whether the target curing pressure is less than the current clamping force. If the clamping force is not less than the current clamping force, control the clamping mechanism to continue increasing the pressure until the target solidification pressure is reached and locking is executed; If the clamping force is less than the current clamping force, an overshoot condition is determined to have violated the monotonically increasing constraint. The pressure is then stopped, and the current clamping force is used as the upper limit of the actual curing pressure. A locking assessment is then performed. Based on the locking assessment result, it is determined whether to use the current clamping force as the curing pressure to perform locking.

8. The method according to claim 1, characterized in that, Before acquiring the excitation current signal of the stacked iron core under constant voltage sinusoidal excitation conditions, the offline construction of pre-configured parameters is also included, specifically including: Using samples from the same batch as the stacked iron core, constant pressure excitation tests were performed under gapless conditions. Controlled compressive stress was applied during the test, and the mapping relationship between stress level and harmonic and fundamental wave characteristics was extracted. Pre-configured single-piece stress harmonic calibration functions and pre-stored gapless magnetization fundamental wave components were constructed respectively. Cyclic loading tests including boost and buck phases were performed on the sample. By comparing the loss response differences between the buck and boost branches, pre-configured hysteresis asymmetric parameters were extracted. Obtain the physical overlap parameters at the joint of the stacked iron core laminations, and extract the pre-configured air gap loss coefficient that matches the physical overlap parameters based on the experience database; Obtain the characteristic parameters of the insulation coating of the stacked iron core and the target curing process state, and determine the pre-configured pressure relaxation parameters before and after curing based on the calculation model.

9. The method according to claim 1, characterized in that, The constant voltage sinusoidal excitation condition is achieved by an AC power supply with low output impedance; the short-circuit impedance of the AC power supply is less than 1% of the excitation impedance of the stacked iron core, so that the external voltage distortion caused by the excitation current harmonics is lower than the preset voltage distortion threshold.

10. A core stacking pressure control system for reducing transformer no-load losses, characterized in that, The system includes a processor coupled to a memory storing instructions which, when executed by the processor, implement the method as claimed in any one of claims 1 to 9.

Images