Method for loss tuning of power transformer

Abstract

The application discloses a loss tuning method of a power transformer, which is applied to a transformer body with a sleeved winding and comprises the following steps: applying a scanning current excitation containing a fundamental wave and a harmonic wave to the winding, synchronously collecting a transformer body vibration signal and converting the transformer body vibration signal into a frequency spectrum; extracting resonance peaks from the frequency spectrum and matching the resonance peaks with a theoretical electromagnetic force wave and a structure inherent frequency library, so that a coupling resonance peak to be tuned is identified; for each resonance peak, an exciter is installed in a corresponding area, an opposite periodic pulse force is emitted, and through real-time monitoring of a vibration response, pulse force parameters are dynamically fine-tuned with the minimum equivalent mechanical impedance as a target; when an optimal damping state is reached, an output rod of the exciter is locked, and a static pre-tightening force is formed; after all the resonance peaks are tuned and the pre-stress is locked in turn, final compression and fixation of the transformer body are completed in a state of keeping the pre-tightening force. The application can actively inhibit and solidify the dynamic loss source of an individual transformer before assembly and solidification, effectively reduce operation loss and noise, and improve structure stability.

Description

Technical Field

[0001] This invention relates to the field of transformer optimization technology, and in particular to a method for optimizing the losses of power transformers. Background Technology

[0002] After a transformer is put into operation, the load current flowing through the windings interacts with the alternating magnetic flux in the core, generating periodic electromagnetic forces at frequencies multiples of the power frequency (such as 100Hz, 200Hz, etc.). These electromagnetic forces act as a continuous "vibration source" on the transformer body, an elastic structure. More problematic is that different materials (such as silicon steel core and copper windings) have different coefficients of thermal expansion, causing them to "compete" with each other during temperature rises due to load changes, generating additional thermal stress. These electromagnetic forces and thermal stresses intertwine, exciting vibrations at specific frequencies in the transformer core, windings, and their fastening structures—a phenomenon known as "structural-electromagnetic coupling resonance." This resonance not only directly causes bothersome electromagnetic noise, but more importantly, it dissipates energy in the form of vibrations and minute deformations. This energy is ultimately converted into additional, uncounted heat, resulting in actual operating losses exceeding theoretical values ​​calculated based on ideal static models. This prevents the transformer from achieving its optimal design efficiency and may even affect reliability due to stress fatigue during long-term operation.

[0003] To address this problem, existing technologies primarily rely on post-event detection and passive optimization. For example, online vibration monitoring systems can detect abnormal vibrations during operation, but this is only an early warning after the problem has occurred, and cannot prevent it during the manufacturing stage. Alternatively, during design and process finalization, finite element simulations can be used to avoid certain obvious resonant frequencies as much as possible, but this is an "avoidance" strategy, and there will always be differences between the simulation model and the actual product, making it impossible to completely eliminate all potential, individualized coupling risks. The fundamental flaw of these methods is that they all fail to proactively intervene and optimize the aforementioned dynamic coupling problem at a core and feasible stage—the window period when the transformer body has been assembled but not yet finalized and secured. As a result, when each transformer leaves the factory, its inherent resonant genes, closely related to individual assembly precision and material properties, are solidified, and its final operational efficiency can only be determined by probability and standard tolerances, making it difficult to guarantee that every product will achieve its theoretically optimal efficiency. Summary of the Invention

[0004] Therefore, the technical problem to be solved by the present invention is to overcome the shortcomings of the prior art in that it is impossible to actively identify and intervene in the individual structural-electromagnetic coupling resonance of the transformer during the manufacturing stage, which is caused by future electromagnetic and thermal loads. This results in actual operating losses exceeding the design value and limited noise and long-term reliability. The present invention provides a loss optimization method for power transformers, which can actively suppress and solidify the root causes of dynamic losses of individual transformers by applying specific excitation, identifying resonance modes, applying reverse counteracting forces, and optimizing and locking static prestress before the final assembly and solidification of the transformer body. This achieves the beneficial effects of effectively reducing actual operating losses, reducing operating noise, and improving structural stability.

[0005] To address the aforementioned technical problems, this invention provides a loss optimization method for power transformers, applicable to transformer bodies after winding assembly, comprising the following steps:

[0006] A set of scanning AC current excitation signals containing the fundamental and harmonic frequencies is applied to the windings of the transformer body; the vibration acceleration signals on the surface of the transformer body are simultaneously acquired and converted into a vibration spectrum.

[0007] From the vibration spectrum, extract multiple resonance peaks whose amplitudes exceed a preset multiple of the ambient noise floor; match the frequency of each resonance peak with a pre-established theoretical electromagnetic force wave frequency library and a structural natural frequency library; mark the successfully matched resonance peaks as resonance peaks to be tuned originating from structure-electromagnetic coupling, and record their frequency fi, phase φi, and vibration-dominant region Ai on the device body;

[0008] For each resonance peak to be tuned, one or more exciters capable of outputting pulse force are installed and fixed on the clamp or pressure plate corresponding to the vibration-dominant region Ai; the exciter is controlled to emit a periodic pulse force with the same frequency fi as the resonance peak to be tuned, but with a phase of φi+180°±Δφ, where Δφ is an adjustable phase angle of less than 30°; the amplitude of the pulse force increases from zero according to a preset slope.

[0009] During the application of pulse force, the vibration response of the vibration-dominant region Ai is monitored in real time, and the equivalent mechanical impedance of the vibration response region is calculated. With the minimization of equivalent mechanical impedance as the control objective, the phase Δφ and amplitude of the pulse force are dynamically fine-tuned. When the equivalent mechanical impedance reaches a local minimum and remains stable for more than a preset time Ts, the state is determined to be the optimal damping state. The extension position of the exciter output rod is kept unchanged at this time, and the exciter is instructed to energize its internal electromagnetic locking mechanism, thereby mechanically locking the output rod at the current displacement. The locked state generates a constant static preload Fi on the device body.

[0010] All marked resonance peaks to be tuned are tuned and prestressed sequentially or in parallel; after all tuning is completed, while keeping all exciters locked, the transformer body is finally tightened and fixed according to the process specifications; then all exciters are released and removed.

[0011] In one embodiment of the present invention, when applying a scanning AC current excitation signal, a staged ramp-up method is adopted, in which the voltage is increased to 30% of the rated voltage with a first voltage slope and maintained for a first duration, and then increased to the target excitation voltage with a second voltage slope that is less than the first voltage slope.

[0012] In one embodiment of the present invention, while simultaneously acquiring the vibration acceleration signal on the surface of the device body, infrared thermal images of key parts of the winding are also acquired simultaneously; the abnormal temperature rise area shown in the infrared thermal image is associated and labeled with the vibration-dominant region Ai identified from the vibration spectrum.

[0013] In one embodiment of the present invention, the pre-established theoretical electromagnetic force wave frequency library is a collection of electromagnetic force harmonic frequencies calculated based on the transformer winding arrangement, ampere-turn distribution and magnetic circuit parameters; the pre-established structural natural frequency library is a collection of its natural frequencies obtained by performing modal tests or finite element modal analysis on the transformer body of the same type.

[0014] In one embodiment of the present invention, before extracting resonance peaks from the vibration spectrum, the acquired vibration acceleration signal is subjected to environmental noise stripping processing. Specifically, before applying excitation, a background vibration signal spectrum of a preset duration is acquired, and the energy of the background vibration signal spectrum is subtracted from the vibration spectrum obtained after applying excitation.

[0015] In one embodiment of the present invention, the initial value of the adjustable phase angle Δφ is set to 0°, and during the dynamic fine-tuning process, it is adaptively adjusted within a range of ±30° according to the gradient of the change in the equivalent mechanical impedance.

[0016] In one embodiment of the present invention, when multiple exciters are installed for the same vibration-dominant region Ai, the periodic pulse forces output by these exciters are controlled to maintain phase synchronization so as to form a composite excitation force superimposed in the same direction.

[0017] In one embodiment of the present invention, the equivalent mechanical impedance of the vibration response region is calculated as follows: at the frequency fi, the ratio of the amplitude of the applied periodic pulse force to the amplitude of the vibration acceleration response measured in the vibration-dominant region Ai is taken as the amplitude of the equivalent mechanical impedance at the frequency point; the difference between the phase of the periodic pulse force and the phase of the vibration acceleration response is taken as the phase of the equivalent mechanical impedance.

[0018] In one embodiment of the present invention, during the final tightening and fixing process, the attenuation of the static preload Fi is monitored in real time; if the attenuation of any static preload Fi exceeds 10% of its set value, the tightening process is paused, and the corresponding vibrator locking status and nearby fasteners are checked and adjusted.

[0019] In one embodiment of the present invention, after releasing and removing all exciters, a scanning AC current excitation signal is applied to the transformer body winding again to verify the optimized and tightened body and detect whether the amplitude of the original resonance peak to be optimized in its vibration spectrum has dropped below the preset qualified threshold.

[0020] The technical solution of the present invention has the following advantages compared with the prior art:

[0021] The power transformer loss optimization method described in this invention introduces an active diagnosis-optimization-fixing system in the final physically intervened stage of transformer manufacturing. Specifically, it first applies a scanning current containing fundamental and harmonic frequencies to the windings to accurately simulate the complex electromagnetic environment the transformer will face in the future, and simultaneously captures the vibration response of the transformer body surface, thereby drawing a dynamic stress diagram (i.e., vibration spectrum) specific to that transformer. Subsequently, by intelligently matching the resonance peaks in the spectrum with theoretical electromagnetic force waves and a library of structural natural frequencies, it can accurately identify which are truly harmful vibration frequencies originating from structural-electromagnetic coupling.

[0022] By installing an exciter in the diagnosed vibration-dominant region and applying a tiny pulse force with the same frequency as the harmful vibration but almost completely opposite in phase (phase difference of 180°±Δφ), a reaction force field is created, directly intervening in and counteracting the generation and transmission path of the harmful vibration energy. Minimizing the equivalent mechanical impedance as the real-time control objective means that the vibration system exhibits the greatest resistance to external excitation at that frequency, thereby fundamentally suppressing the energy level of the resonance peak.

[0023] When the optimal damping state is achieved, the system instructs the exciter to lock the position of its output rod, thereby converting the instantaneous mechanical force that produces this perfect counteracting effect into a constant, static preload, permanently applied to the corresponding structure of the transformer body. Finally, in this optimized state where all prestresses have been applied, the final mechanical fastening of the transformer body is completed.

[0024] Based on the above principles, the beneficial effects of this technical solution include:

[0025] 1. By actively suppressing specific frequency vibrations caused by structure-electromagnetic coupling, the additional core magnetostriction losses, winding eddy current losses, and mechanical friction losses caused by such vibrations are directly reduced. This reduces the total losses of the transformer in actual operation, and the operating efficiency is closer to its theoretical limit based on ideal materials and design.

[0026] 2. Because harmful electromagnetic vibration modes are effectively suppressed at the manufacturing source, the transformer's inherent noise level under rated and specific harmonic loads will be significantly reduced.

[0027] 3. The optimized prestress state solidified inside the transformer body can partially offset the periodic dynamic stress generated by load changes and electromagnetic forces during future operation, thereby reducing stress fatigue of key insulation and fastening components, helping to extend the service life of the transformer and improve its operational reliability. Attached Figure Description

[0028] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein:

[0029] Figure 1 This is a flowchart of the steps of the power transformer loss optimization method of the present invention;

[0030] Figure 2 This is a flowchart of the steps for multimodal synchronous monitoring and identification of vibration-dominant regions according to the present invention;

[0031] Figure 3 This is a flowchart of the phase angle adaptive adjustment steps of the present invention;

[0032] Figure 4 This is a flowchart of the steps for calculating the equivalent mechanical impedance of the present invention. Detailed Implementation

[0033] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0034] Reference Figure 1As shown, this invention discloses a loss optimization method for power transformers, applied to the transformer body after the windings have been assembled. A power transformer is an electrical device that uses the principle of electromagnetic induction to change AC voltage. During operation, it generates electromagnetic vibrations and losses. This method is applied during the manufacturing process of power transformers, specifically at the stage where the windings have been assembled but the transformer body has not yet been finally tightened and fixed. The transformer body after winding assembly refers to a semi-finished state where the transformer core, windings, insulation components, and other main components have been assembled according to design requirements, but have not yet undergone final tightening, fixing, or tank sealing. At this stage, the transformer body structure has a certain degree of adjustability, providing operational flexibility for the implementation of this method.

[0035] A loss optimization method for a power transformer according to the present invention includes the following steps: applying a set of scanning AC current excitation signals containing fundamental and harmonic frequencies to the windings of the transformer body; synchronously acquiring vibration acceleration signals on the surface of the transformer body and converting them into a vibration spectrum; extracting multiple resonance peaks from the vibration spectrum whose amplitudes exceed a preset multiple of the ambient noise floor; matching the frequency of each resonance peak with a pre-established theoretical electromagnetic force wave frequency library and a structural natural frequency library; marking the successfully matched resonance peaks as resonance peaks to be optimized originating from structure-electromagnetic coupling, and recording their frequency fi, phase φi, and vibration-dominant region Ai on the transformer body; for each resonance peak to be optimized, installing and fixing one or more exciters capable of outputting pulse forces on the transformer body clamps or pressure plates corresponding to the vibration-dominant region Ai; controlling the exciters to emit periodic pulse forces with the same frequency fi as the resonance peak to be optimized but with a phase of φi+180°±Δφ, where Δφ is a periodic pulse force. The adjustable phase angle is less than 30°; the amplitude of the pulse force increases from zero at a preset slope; during the application of the pulse force, the vibration response of the dominant vibration region Ai is monitored in real time, and the equivalent mechanical impedance of the vibration response region is calculated; with the minimization of the equivalent mechanical impedance as the control objective, the phase Δφ and amplitude of the pulse force are dynamically fine-tuned; when the equivalent mechanical impedance reaches a local minimum and remains stable for more than a preset time Ts, the state is determined to be the optimal damping state; the extension position of the exciter output rod remains unchanged at this time, and the exciter is instructed to energize its internal electromagnetic locking mechanism, thereby mechanically locking the output rod at the current displacement, and the locked state generates a constant static preload Fi on the transformer body; all marked resonance peaks to be tuned are tuned and prestressed locked sequentially or in parallel; after all tuning is completed, while keeping all exciters locked, the transformer body is finally pressed and fixed according to the process specifications; then all exciters are released and removed.

[0036] Among them, the scanning AC current excitation signal refers to an AC current signal whose frequency and / or amplitude continuously vary within a certain range. It is applied to the transformer winding to simulate the electromagnetic force that the transformer is subjected to in actual operation, thereby stimulating the potential vibration response in the transformer body structure for detection and analysis.

[0037] The vibration spectrum is obtained by processing the collected vibration acceleration signal through Fourier transform and other methods. It represents the distribution of vibration energy at different frequencies. By analyzing the vibration spectrum, the resonance phenomenon of the structure at a specific frequency can be identified. The resonance peak refers to the peak value in the vibration spectrum where the vibration amplitude is significantly higher than the surrounding frequency points. These peak values ​​usually correspond to the resonance of the structure at a specific frequency, indicating that the structure responds particularly strongly to the excitation at that frequency.

[0038] A vibrator is a device that can generate periodic mechanical force. In this method, the vibrator is used to apply a pulsed force with the same and controllable phase as the resonant peak frequency fi to be tuned to the transformer body in order to actively intervene in and suppress structural vibration.

[0039] Equivalent mechanical impedance is a physical quantity that measures a structure's resistance to vibrational responses caused by external excitations. The smaller the value, the weaker the structure's resistance to vibration at that frequency, and the more prone it is to resonance. This method aims to minimize the equivalent mechanical impedance, thereby achieving optimal damping of the structure's vibrational response at a specific frequency through external intervention.

[0040] Static preload Fi refers to a constant mechanical force generated and maintained on the vibrator structure by locking the output rod at its current displacement through its internal electromagnetic locking mechanism after the vibrator reaches its optimal damping state. This preload aims to change the natural frequency and damping characteristics of the vibrator structure in order to suppress the original structure-electromagnetic coupling resonance.

[0041] In implementing this method, a set of scanning AC current excitation signals containing the fundamental and harmonic frequencies is first applied to the windings of the transformer body. For example, an adjustable power supply can be used to gradually increase the voltage at a constant slope, causing the current to increase linearly from zero to the target value, while continuously scanning the frequency during this process. Simultaneously, multiple accelerometers are placed on the surface of the transformer body to synchronously record its vibration acceleration data. These raw data are then subjected to Fourier transform by a digital signal processor to generate a vibration spectrum.

[0042] Furthermore, resonance peaks are extracted from the obtained vibration spectrum. Specifically, by setting a fixed vibration amplitude threshold, frequency points in the vibration spectrum exceeding this threshold are identified as resonance peaks. Subsequently, the frequency of each resonance peak is matched with a pre-established theoretical electromagnetic force wave frequency library and a structural natural frequency library. The theoretical electromagnetic force wave frequency library can estimate the main harmonic frequencies based on the general electromagnetic force calculation formula provided in the transformer design manual. The structural natural frequency library can be obtained by empirically testing transformer bodies of similar size and structure to obtain their approximate natural frequency range. Successfully matched resonance peaks are marked as resonance peaks to be tuned originating from structure-electromagnetic coupling, and their frequency fi, phase φi, and vibration-dominant region Ai on the transformer body are recorded. The vibration-dominant region Ai can be identified by densely arranging sensors on the transformer body surface or by manually scanning sensors, finding the area with the largest vibration amplitude at a specific resonance frequency, and marking it.

[0043] For each resonance peak to be tuned, one or more exciters capable of outputting pulse forces are mounted and fixed on the clamping parts or pressure plates corresponding to the dominant vibration region Ai. The exciters can be set to output periodic pulse forces with the same frequency fi as the resonance peak to be tuned. The initial phase can be set to φi+180°, and Δφ can be adjusted manually or through a simple stepping algorithm. The amplitude of the pulse force can be manually controlled by the operator, gradually increasing according to a preset fixed slope.

[0044] During the application of the pulsed force, the vibration response of the dominant vibration region Ai is monitored in real time, and the equivalent mechanical impedance of the vibration response region is calculated. The evaluation of the equivalent mechanical impedance can be simplified to directly observing the change in the vibration acceleration response amplitude. When the vibration amplitude reaches its minimum value, the equivalent mechanical impedance is considered to have reached a local minimum. With minimizing the equivalent mechanical impedance as the control objective, the phase Δφ and amplitude of the pulsed force can be adjusted manually or through a simple feedback loop. For example, the operator can manually adjust Δφ and amplitude, observing the change in the vibration response amplitude until the vibration amplitude no longer decreases significantly. When the vibration response amplitude reaches a local minimum, and this minimum value does not fluctuate significantly within a time period Ts, the optimal damping state can be determined.

[0045] Once the optimal damping state is reached, the extended position of the exciter output rod remains unchanged, and the exciter is instructed to energize its internal electromagnetic locking mechanism, thereby mechanically locking the output rod at the current displacement. This locked state generates a constant static preload Fi on the transformer body. The above optimization and locking operations can be performed on all identified resonance peaks to be optimized one by one, or, if conditions permit, multiple resonance peaks can be optimized simultaneously. After all optimizations are completed, with all exciters locked and preload applied, the transformer body is finally mechanically pressed and fixed according to the standard transformer manufacturing process. Subsequently, the exciter's locking state is released, and it is removed from the transformer body surface.

[0046] This method actively applies excitation and identifies structural-electromagnetic coupling resonance during the critical window period after the transformer windings are assembled and before final tightening. Dynamic optimization is achieved by applying a counter-phase pulse force through a vibrator, ultimately locking in a static preload. This effectively alters the inherent characteristics of the transformer structure and suppresses resonance. Consequently, it reduces energy dissipation during transformer operation at the source, lowers actual operating losses, improves transformer efficiency and long-term reliability, and solves the problem that traditional methods cannot actively intervene in resonance during the manufacturing stage.

[0047] Specifically, the above embodiments of this application propose applying a set of scanning AC current excitation signals containing fundamental and harmonic frequencies to the windings of the transformer body, and simultaneously acquiring vibration acceleration signals from the transformer body surface for subsequent analysis. However, in actual operation, if the voltage application method of the excitation signal is inappropriate, such as the voltage rising too quickly or too abruptly, it may cause unnecessary mechanical shock or thermal stress to the transformer body after the windings have been fitted, especially when the internal structure of the transformer has not yet fully adapted to the excitation state. This inappropriate excitation method may cause the measured vibration response to contain transient components, thereby affecting the accuracy of resonance peak extraction, and may even cause potential damage to the transformer.

[0048] In response, this application further proposes that when applying a scanning AC current excitation signal, a staged ramp-up voltage method is adopted, in which the voltage is increased to 30% of the rated voltage with a first voltage slope and maintained for a first duration, and then increased to the target excitation voltage with a second voltage slope that is less than the first voltage slope.

[0049] Specifically, the phased ramp voltage boosting method refers to gradually increasing the voltage of the excitation signal from zero to the target value, and this boosting process is divided into at least two stages, each using a different voltage rise rate (slope). This phased, gradual voltage application method aims to avoid applying instantaneous or excessively rapid voltage surges to the transformer body, thereby protecting the internal structure of the transformer and ensuring accurate vibration response data is obtained under stable conditions.

[0050] In the first stage, the voltage rises from zero at a first voltage slope until it reaches 30% of the rated voltage. For example, the first voltage slope can be set to 10 to 50 volts per second. Once this voltage level is reached, the system maintains this voltage for a preset first duration, such as 5 to 30 seconds. The purpose of this stage is to provide a gentle pre-adaptation process for the transformer. With a lower voltage and a controlled rate of rise, the transformer's magnetic circuit is gradually established, the windings initially stabilize under moderate electromagnetic forces, and any initial transient response is allowed to decay. This helps eliminate nonlinear effects caused by sudden voltage increases, providing a stable reference state for subsequent higher voltage excitation.

[0051] Subsequently, in the second stage, the voltage will continue to rise at a second voltage slope, less than the first voltage slope, until the final target excitation voltage is reached. For example, if the first voltage slope is 20V / s, the second voltage slope can be set to 5V / s to 15V / s. The target excitation voltage can be the transformer's rated voltage or other specific voltage values ​​set according to testing requirements. Using a smaller second voltage slope is to further slow the voltage rise rate in areas with higher voltage and stronger electromagnetic forces. This is because as voltage and current increase, the electromagnetic forces inside the transformer and their impact on the mechanical structure significantly increase. By reducing the voltage rise rate in this stage, the risk of transient mechanical shocks and localized overheating that may occur under high voltage can be minimized, ensuring that the transformer structure responds to the excitation more smoothly and linearly, thereby obtaining more reliable vibration data.

[0052] Specifically, the above embodiments of this application propose that by analyzing the vibration acceleration signal on the surface of the transformer body, the dominant vibration region Ai can be identified and optimized. However, the internal losses of the transformer are not only reflected in mechanical vibration, but may also be accompanied by phenomena such as local overheating. If these hot spots are not effectively identified and dealt with, the optimization effect may be incomplete, or even potential fault hazards may be missed.

[0053] In this regard, refer to Figure 2 As shown, this application further proposes a multimodal synchronous monitoring technical solution, which simultaneously acquires the vibration acceleration signal on the surface of the device body and the infrared thermal image of the key parts of the winding; the abnormal temperature rise area shown in the infrared thermal image is associated and labeled with the vibration dominant area Ai identified from the vibration spectrum.

[0054] Specifically, synchronous acquisition of infrared thermal images of key winding components refers to the non-contact temperature field measurement of specific key parts of the transformer windings using an infrared thermal imager during the process of applying a scanning AC current excitation signal containing fundamental and harmonic frequencies to the transformer body and simultaneously acquiring vibration acceleration signals. These key parts typically include the winding ends, lead connections, tap changer areas, and areas where partial discharge or insulation defects may exist. These areas are common points of internal losses and faults in transformers. The infrared thermal imager should have sufficient spatial resolution and temperature measurement accuracy to accurately capture subtle temperature changes. The acquired infrared thermal images should be synchronized with the vibration acceleration signals in time or have precise timestamps for subsequent data correlation.

[0055] The association and annotation of abnormal temperature rise areas displayed in the infrared thermal image with the vibration-dominant region Ai identified from the vibration spectrum refers to the process of identifying local areas with temperatures higher than the normal operating temperature threshold or the ambient temperature after acquiring the infrared thermal image, using image processing and analysis techniques. These areas are determined to be abnormal temperature rise areas. The determination of abnormal temperature rise can be based on a preset temperature threshold, comparison with a reference temperature field, or temperature gradient analysis. Subsequently, the spatial location information of these abnormal temperature rise areas on the transformer body is compared and mapped with the spatial location information of the vibration-dominant region Ai identified through vibration spectrum analysis. If an abnormal temperature rise area overlaps or is closely adjacent to a vibration-dominant region Ai in space, the vibration-dominant region Ai is associated and annotated, indicating that it is not only a region of active mechanical vibration but also a region with thermal anomalies. This association and annotation can be achieved by adding markers to data records or highlighting them on a visualization interface.

[0056] Specifically, this application further clarifies the construction methods of the pre-established theoretical electromagnetic force wave frequency library and the structural natural frequency library. The pre-established theoretical electromagnetic force wave frequency library is a collection of electromagnetic force harmonic frequencies calculated based on the transformer winding arrangement, ampere-turn distribution, and magnetic circuit parameters; the pre-established structural natural frequency library is a collection of its natural frequencies obtained by modal testing or finite element modal analysis of the transformer body of the same type.

[0057] Specifically, the pre-established theoretical electromagnetic force wave frequency library aims to provide theoretical predictions of the frequencies of electromagnetic force vibrations that may be generated inside the transformer. Its construction is based on the actual arrangement of the transformer windings, such as concentric, interleaved, disc, or spiral structures, as different winding structures affect the magnetic field distribution. Simultaneously, the ampere-turn distribution, i.e., the distribution of the product of current and number of turns along the winding axis or radial direction, is a key input for calculating the electromagnetic force. Furthermore, the transformer's magnetic circuit parameters, such as the permeability of the core and the geometry of the leakage flux channels, directly affect the calculation of magnetic field strength and electromagnetic force. By comprehensively considering these parameters and applying electromagnetic field theory (such as Maxwell's equations) and mechanical principles (such as Lorentz force or Maxwell's stress tensor method), the frequencies of the fundamental and harmonic electromagnetic forces generated inside the transformer at different excitation frequencies can be accurately calculated. These calculation results constitute a set of theoretical electromagnetic force wave frequencies, providing a reliable theoretical basis for subsequent resonance peak matching.

[0058] A pre-established library of structural natural frequencies is used to characterize the vibration characteristics of the transformer core structure. This library is obtained in two ways: one is through modal testing, conducted on a transformer core of the same type as the transformer to be optimized, under non-energized conditions. Excitation is applied to the core using a vibrator (such as a hammer or electric vibrator), and the vibration response at various points on the core is collected using an accelerometer. Modal analysis of the collected data identifies the natural frequencies and damping ratios of the core under different vibration modes. The other method is through finite element modal analysis, establishing a precise three-dimensional finite element model of the transformer core, including key structural components such as the core, windings, clamps, pressure plates, and tank, and assigning them corresponding material properties (such as elastic modulus, density, and Poisson's ratio) and boundary conditions. Modal analysis calculations using finite element software yield the natural frequencies and corresponding mode shapes of the core. Both methods provide a set of natural frequencies for the core structure under different vibration modes, providing a structural reference for identifying structure-electromagnetic coupling resonances.

[0059] By precisely constructing the theoretical electromagnetic force wave frequency library and the structural natural frequency library as described above, this application can significantly improve the accuracy and reliability of resonance peak matching. The electromagnetic force harmonic frequencies calculated based on the transformer winding arrangement, ampere-turn distribution, and magnetic circuit parameters ensure that the identified electromagnetic force source has a clear physical basis. Simultaneously, the structural natural frequencies obtained through modal testing or finite element analysis accurately reflect the vibration characteristics of the transformer itself. This dual-precise frequency library matching mechanism enables the system to more accurately distinguish vibrations truly caused by the coupling of electromagnetic force and structural natural frequencies, rather than merely environmental noise or pure structural vibration.

[0060] The power transformer loss optimization method described in this application faces challenges when performing vibration spectrum analysis on the transformer body to identify resonance peaks. Industrial environments often contain various background noises, such as those from surrounding machinery, power grid fluctuations, or environmental vibrations. These noises may overlap with the transformer's own vibration signals, leading to inaccurate resonance peaks extracted from the vibration spectrum. Furthermore, noise-induced peaks may be misidentified as structural-electromagnetic coupling resonances, thus affecting the accuracy and effectiveness of subsequent optimization.

[0061] To this end, this application further proposes to perform environmental noise stripping on the acquired vibration acceleration signal before extracting resonance peaks from the vibration spectrum. Environmental noise stripping is a signal processing technique designed to remove or reduce unwanted noise components from the measured signal, thereby improving the signal-to-noise ratio and enhancing the clarity and accuracy of the desired signal characteristics. In vibration analysis, environmental noise stripping is crucial for distinguishing true structural resonances caused by electromagnetic forces from stray peaks or elevated noise floors introduced by the surrounding environment. Its purpose is to ensure that the extracted resonance peaks truly represent the internal vibration characteristics of the transformer, rather than external interference.

[0062] Specifically, before applying excitation, a background vibration signal spectrum of a preset duration is acquired. This step aims to characterize the environmental noise present in the measurement environment before the transformer itself is actively excited. By capturing noise characteristics without the influence of an excitation signal, a baseline of environmental interference can be established. The preset duration should be long enough to capture a representative sample of the environmental noise and account for its potential temporal variations, but not so long as to excessively prolong the measurement process. Typically, this duration can range from a few seconds to several minutes, depending on the stability of the noise environment. The background vibration signal spectrum is obtained by performing a Fast Fourier Transform (FFT) or similar spectral analysis on the time-domain vibration data acquired during this period, which provides a frequency domain representation of the environmental noise energy.

[0063] Subsequently, the energy of the background vibration signal spectrum is subtracted from the vibration spectrum obtained after subsequent excitation. By subtracting the pre-measured noise spectrum from the spectrum obtained during excitation, the contribution of environmental noise to the overall measured vibration is effectively removed or significantly reduced. This leaves a clearer spectrum that more accurately reflects the transformer's response to the applied excitation. "Subtracted accordingly" means subtraction on a frequency-by-frequency basis. For each frequency component in the spectrum obtained during excitation, the corresponding energy (or the square of the amplitude) in the background noise spectrum is subtracted. This assumes that the environmental noise characteristics remain relatively constant between the background measurement and the excitation measurement.

[0064] Reference Figure 3As shown, this application further proposes that the initial value of the adjustable phase angle Δφ be set to 0°, and during the dynamic fine-tuning process, adaptive adjustment is performed within a range of ±30° based on the change gradient of the equivalent mechanical impedance.

[0065] Specifically, when tuning each resonance peak to be tuned, the periodic pulse force output by the exciter is designed to superimpose in opposite phase with the inherent vibration of the dominant vibration region Ai of the vibrator body, thereby suppressing vibration. Theoretically, the ideal phase difference for this anti-phase superposition is 180°. Therefore, setting the initial value of the adjustable phase angle Δφ to 0° means that at the start of dynamic fine-tuning, the phase of the pulse force output by the exciter is φi + 180°. This provides a clear and central starting point, facilitating subsequent fine-tuning operations and ensuring that the tuning process begins from a state close to the ideal anti-phase state.

[0066] The gradient of the equivalent mechanical impedance refers to the rate and direction of change of the equivalent mechanical impedance with respect to the adjustable phase angle Δφ. During dynamic fine-tuning, the system periodically and slightly changes Δφ and monitors the resulting changes in the equivalent mechanical impedance. By calculating these changes, the gradient information at the current Δφ point can be obtained. For example, if increasing Δφ leads to a decrease in impedance, the gradient is negative, indicating that adjustment should continue in the direction of increasing Δφ; if decreasing Δφ leads to a decrease in impedance, the gradient is positive, indicating that adjustment should continue in the direction of decreasing Δφ. "Adaptive adjustment" refers to intelligently adjusting the step size and direction of Δφ based on the calculated gradient. When the gradient is large, a larger step size can be used to accelerate convergence; when the gradient is small, a smaller step size is used to avoid overshoot and improve adjustment accuracy. This adaptive mechanism ensures that the tuning process approximates the local minimum of the equivalent mechanical impedance quickly and accurately. The search space of the adjustable phase angle Δφ is defined "within ±30°". This range is set based on empirical or theoretical analysis to ensure that the tuning process is carried out within a reasonable physical range, avoiding excessive phase deviation between the pulse force output by the exciter and the vibration of the transformer body, thereby ensuring the effectiveness of the tuning. This range also takes into account the complexity of the actual transformer structure and the effect of electromagnetic forces, allowing for a certain degree of deviation to find the optimal anti-phase suppression effect.

[0067] Furthermore, when multiple exciters are installed for the same dominant vibration region Ai, the periodic pulse forces output by these exciters are controlled to maintain phase synchronization, so as to form a composite excitation force superimposed in the same direction.

[0068] Specifically, during loss optimization of power transformer bodies, when a dominant vibration region Ai is large or its internal vibration energy is significant, the pulse force provided by a single exciter may be insufficient to effectively suppress resonance in that region. In this case, to enhance vibration suppression in that region, multiple exciters can be strategically installed and deployed within or around the dominant vibration region Ai. These exciters can be distributed at different locations within the dominant vibration region Ai, based on the characteristics of the vibration mode—for example, uniformly distributed, concentrated at the point of most intense vibration, or arranged along a specific direction—to ensure effective coverage and action across the entire region. Phase synchronization refers to the consistent phase relationship in time between the periodic pulse forces output by all exciters within the same dominant vibration region Ai. This means that when the pulse force output by one exciter reaches its peak, other exciters simultaneously output their peak pulse forces; conversely, when the pulse force output by one exciter is at its trough, other exciters are also at their troughs. This synchronous control can be achieved through a central controller that receives the frequency fi and phase φi of the resonance peak to be tuned and sends synchronous control commands to all exciters, ensuring that they output pulse forces according to a preset phase relationship (i.e., φi + 180° ± Δφ). Co-directional superposition refers to the fact that the pulse forces output by multiple exciters are consistent in their direction of action, and due to phase synchronization, these forces can be constructively superimposed. For example, if all exciters are designed to apply thrust to the interior of the vibrator, then under phase synchronization, their thrust will act simultaneously on the vibrator, forming a composite pulse force that is much larger, more directional, and has a greater amplitude than that provided by a single exciter. This composite excitation force can more effectively counteract the resonance of the dominant vibration region Ai, thereby significantly reducing the vibration response in that region.

[0069] Reference Figure 4 As shown, this application further proposes a specific method for calculating the equivalent mechanical impedance of the vibration response region. At the frequency fi, the ratio of the amplitude of the applied periodic pulse force to the amplitude of the vibration acceleration response measured in the vibration-dominant region Ai is taken as the amplitude of the equivalent mechanical impedance at that frequency point; the difference between the phase of the periodic pulse force and the phase of the vibration acceleration response is taken as the phase of the equivalent mechanical impedance.

[0070] Regarding the calculation of the equivalent mechanical impedance amplitude, this application proposes to define it as the ratio of the amplitude of the periodic pulse force applied to the vibration-dominant region Ai at a specific frequency fi to the amplitude of the vibration acceleration response measured in that region. Specifically, the amplitude of the periodic pulse force can be monitored in real time by installing a force sensor at the output of the exciter, while an acceleration sensor is attached to the surface of the vibration-dominant region Ai to obtain the amplitude of the vibration acceleration response. By performing Fourier transform or other spectral analysis on these two signals, the amplitude information at frequency fi is extracted, and their ratio is calculated. This ratio intuitively reflects the structure's resistance to external excitation at a specific frequency; the smaller the value, the easier the structure is to be excited at that frequency.

[0071] For the phase calculation of equivalent mechanical impedance, this application proposes to define it as the difference between the phase of the periodic pulse force and the phase of the vibration acceleration response. After acquiring the periodic pulse force signal and the vibration acceleration response signal, the phase information of these two signals at frequency fi is extracted using spectral analysis methods (e.g., Fast Fourier Transform). Then, the difference between these two phases is calculated. This phase difference provides important information about the dynamic response of the structure. For example, a phase difference close to 180° may indicate that the structure is in a resonant state, while a phase difference close to 0° may indicate that the structural response is in phase with the excitation.

[0072] The above technical solution clarifies the calculation method for equivalent mechanical impedance. The amplitude is determined by the ratio of the applied periodic pulse force amplitude to the vibration acceleration response amplitude measured in the dominant vibration region Ai, and the phase is determined by the difference between the phase of the periodic pulse force and the phase of the vibration acceleration response. This explicit calculation method allows for accurate and real-time acquisition of complete information on the equivalent mechanical impedance during optimization, providing a precise control basis for the dynamic fine-tuning of the exciter's pulse force. Based on this, the system can more effectively identify and track the local minimum value of the equivalent mechanical impedance, ensuring that the transformer body reaches the optimal damping state, significantly improving the accuracy and efficiency of vibration suppression, and thus effectively reducing the losses of the power transformer.

[0073] After all the tuning is completed, when the transformer body is finally tightened and fixed, the external tightening operation may interfere with the static preload Fi that has been established in the vibrator, causing the preload to decay or fail, which in turn affects the achieved optimal damping state and makes it impossible to effectively maintain the tuning effect.

[0074] In this regard, this application further proposes to monitor the attenuation of static preload Fi in real time during the final tightening and fixing process; if the attenuation of any static preload Fi exceeds 10% of its set value, the tightening process is suspended, and the locking status of the corresponding vibrator and nearby fasteners are checked and adjusted.

[0075] Specifically, during the final clamping and fixing of the transformer body, it is necessary to continuously acquire the current value of the static preload Fi applied by each exciter. This can be achieved by integrating miniature force sensors (e.g., piezoelectric force sensors or strain gauge force sensors) onto the exciter output rod. These sensors can convert force signals into electrical signals and transmit them to the monitoring system. Alternatively, thin-film pressure sensors can be installed on the contact surfaces between the exciter and the body clamps or pressure plates to directly measure local pressure changes. The monitoring system periodically collects data from these force sensors and compares it with the static preload Fi setpoint recorded at the initial moment of exciter locking. The monitoring system compares the real-time acquired static preload Fi value with the initial setpoint to calculate the current attenuation percentage. For example, if the initial setpoint is Fset and the real-time monitoring value is Fcurrent, the attenuation percentage is (Fset - Fcurrent) / Fset * 100%. When this calculated attenuation percentage exceeds a preset 10% threshold, the system triggers a corresponding anomaly handling mechanism. This 10% threshold is set based on engineering experience and the requirements for the structural stability of transformers. It aims to allow for normal fluctuations within a certain range while promptly detecting significant attenuations that may affect the tuning effect.

[0076] Once the system detects that the attenuation of any static preload Fi exceeds a preset threshold, it immediately sends a command to the ongoing final tightening and fixing equipment to stop the current operation. For example, if an automated hydraulic tightening device is used, it will cut off the power to the hydraulic pump or close the control valve; if an electric torque wrench is used for bolt tightening, it will stop its operation. For manual operation, the system will clearly inform the operator to immediately stop the tightening work through audible and visual alarms and display prompts to prevent further deterioration of the preload or irreversible damage. After the tightening process is paused, the specific cause of the preload attenuation needs to be investigated. First, check the vibrator that is experiencing attenuation to confirm whether its internal electromagnetic locking mechanism is still in a valid locked state and whether the output rod has been unexpectedly displaced or loosened. If the vibrator itself has a problem, it may need to be re-energized and locked or replaced. Second, check the fasteners such as the clamps, pressure plates, and connecting bolts near the vibrator's installation location. Loosening, deformation, shim failure, or improper installation of these components may all cause the preload to be unable to be effectively transmitted or maintained. The inspection includes checking the bolt tightening torque, component clearance, and the presence of cracks or plastic deformation. Based on the inspection results, appropriate adjustments are made, such as retightening bolts, replacing damaged shims, adjusting component positions, or relocking the vibrator until the static preload Fi returns to the set value or an acceptable range.

[0077] By introducing a real-time monitoring mechanism for the static preload Fi during the final tightening and fixing process, and setting an attenuation threshold, this application can promptly detect and intervene in preload failure issues that may arise from subsequent tightening operations. When the attenuation of the preload Fi exceeds the set value, the tightening process is immediately paused, and the vibrator's locking status and nearby fasteners are inspected and adjusted. This effectively avoids the risk of the optimal damping state carefully established by the vibrator being destroyed during the final assembly stage. This ensures the long-term stability and reliability of transformer loss optimization, allowing the prestress applied by the vibrator to continue to play an effective role, thereby maintaining the transformer's low vibration and low loss characteristics during operation, significantly improving the overall effect of the optimization method and the transformer's operating performance.

[0078] After removing all exciters, a scanning AC current excitation signal is applied again to the transformer windings to verify the optimized and tightened transformer body. Specifically, after optimizing and locking the prestress of all resonant peaks to be optimized in the transformer body, and performing final clamping and fixing of the transformer body according to the process specifications, the exciters used to apply pulse force and generate static prestress have been removed from the transformer body. At this point, the transformer body is in its final assembled state, and its internal prestressed structure has been maintained by the locking of the exciters and subsequent mechanical fastening. Based on this, an AC current signal with controllable frequency and amplitude is again applied to the transformer windings using a dedicated excitation power supply or testing equipment. This signal is typically scanned from low to high frequency to cover the entire frequency range in which the transformer may generate electromagnetic vibrations, and includes fundamental and harmonic components to simulate the electromagnetic force characteristics in actual operation. This step aims to re-excite the electromagnetic vibration characteristics exhibited by the transformer body before optimization for comparative analysis.

[0079] While the scanning AC current excitation signal is applied again, the vibration acceleration signal on the surface of the transformer body is simultaneously acquired and converted into a vibration spectrum. Then, the amplitude of the original resonant peak to be tuned in the vibration spectrum is detected to determine whether it has decreased below the preset acceptable threshold. The purpose of this step is to finally confirm the effectiveness of the entire tuning process. It not only verifies whether the prestress applied by the exciter has successfully reduced vibration, but also considers whether the tuning effect remains stable and effective after the exciter is removed and after subsequent mechanical operations such as final clamping and fixing of the transformer body. Specifically, the focus is on the frequency points corresponding to the resonant peaks to be tuned identified before tuning. By analyzing the vibration amplitude at these specific frequency points, the change in vibration energy before and after tuning can be quantitatively evaluated. A preset acceptable threshold, such as the maximum allowable vibration amplitude set according to factors such as transformer type, capacity, operating environment, and customer requirements, will serve as the standard for judging the success of the tuning effect. If, after tuning and tightening, the amplitude of the original resonant peak to be tuned in the vibration spectrum of the transformer body is lower than or equal to this threshold, the tuning is considered successful, and the transformer body meets the vibration performance requirements.

[0080] This application provides an objective and quantitative method to verify the final effect of power transformer loss optimization by applying a scanning AC current excitation signal to the transformer windings again after the exciter has been removed and the transformer body has undergone final tightening and fixing, and then detecting the original resonance peak amplitude in its vibration spectrum. This method can directly assess whether the original structural-electromagnetic coupling resonance peak amplitude of the transformer body, after prestressing locking and final tightening, has been effectively reduced and meets the preset qualification standard. This not only ensures the effectiveness of the optimization process and avoids the attenuation of the optimization effect that may be caused by the removal of the exciter or subsequent tightening operations, but also provides a reliable basis for quality control before the transformer leaves the factory, thereby ensuring that the transformer has lower vibration and losses in actual operation and extending its service life.

[0081] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A loss optimization method for a power transformer, applied to the transformer body after winding assembly, characterized in that, Includes the following steps: A set of scanning AC current excitation signals containing the fundamental and harmonic frequencies is applied to the windings of the transformer body; the vibration acceleration signals on the surface of the transformer body are simultaneously acquired and converted into a vibration spectrum. From the vibration spectrum, extract multiple resonance peaks whose amplitudes exceed a preset multiple of the ambient noise floor; match the frequency of each resonance peak with a pre-established theoretical electromagnetic force wave frequency library and a structural natural frequency library; mark the successfully matched resonance peaks as resonance peaks to be tuned originating from structure-electromagnetic coupling, and record their frequency fi, phase φi, and vibration-dominant region Ai on the device body; For each resonance peak to be tuned, one or more exciters capable of outputting pulse force are installed and fixed on the clamp or pressure plate corresponding to the vibration-dominant region Ai; the exciter is controlled to emit a periodic pulse force with the same frequency fi as the resonance peak to be tuned, but with a phase of φi+180°±Δφ, where Δφ is an adjustable phase angle of less than 30°; the amplitude of the pulse force increases from zero according to a preset slope. During the application of pulse force, the vibration response of the vibration-dominant region Ai is monitored in real time, and the equivalent mechanical impedance of the vibration response region is calculated. With the minimization of equivalent mechanical impedance as the control objective, the phase Δφ and amplitude of the pulse force are dynamically fine-tuned. When the equivalent mechanical impedance reaches a local minimum and remains stable for more than a preset time Ts, the state is determined to be the optimal damping state. The extension position of the exciter output rod is kept unchanged at this time, and the exciter is instructed to energize its internal electromagnetic locking mechanism, thereby mechanically locking the output rod at the current displacement. The locked state generates a constant static preload Fi on the device body. All marked resonance peaks to be tuned are tuned and prestressed sequentially or in parallel; after all tuning is completed, while keeping all exciters locked, the transformer body is finally tightened and fixed according to the process specifications; then all exciters are released and removed.

2. The loss optimization method for power transformers according to claim 1, characterized in that: When applying the scanning AC current excitation signal, a staged ramp-up method is adopted. The voltage is increased to 30% of the rated voltage with a first voltage slope and maintained for a first duration, and then increased to the target excitation voltage with a second voltage slope that is less than the first voltage slope.

3. The loss optimization method for power transformers according to claim 1, characterized in that: While synchronously acquiring vibration acceleration signals on the surface of the device body, infrared thermal images of key parts of the winding are also acquired simultaneously; the abnormal temperature rise areas shown in the infrared thermal images are associated and labeled with the vibration-dominant region Ai identified from the vibration spectrum.

4. The loss optimization method for power transformers according to claim 1, characterized in that: The pre-established theoretical electromagnetic force wave frequency library is a collection of electromagnetic force harmonic frequencies calculated based on the transformer winding arrangement, ampere-turn distribution, and magnetic circuit parameters; the pre-established structural natural frequency library is a collection of its natural frequencies obtained by performing modal tests or finite element modal analysis on the transformer body of the same model.

5. The loss optimization method for power transformers according to claim 1, characterized in that: Before extracting resonance peaks from the vibration spectrum, the acquired vibration acceleration signal is subjected to environmental noise stripping. Specifically, before applying excitation, a background vibration signal spectrum of a preset duration is acquired, and the energy of the background vibration signal spectrum is subtracted from the vibration spectrum obtained after applying excitation.

6. The loss optimization method for power transformers according to claim 1, characterized in that: The initial value of the adjustable phase angle Δφ is set to 0°. During the dynamic fine-tuning process, it is adaptively adjusted within a range of ±30° according to the change gradient of the equivalent mechanical impedance.

7. The loss optimization method for power transformers according to claim 1, characterized in that: When multiple exciters are installed for the same dominant vibration region Ai, the periodic pulse forces output by these exciters are controlled to maintain phase synchronization in order to form a composite excitation force that is superimposed in the same direction.

8. The loss optimization method for power transformers according to claim 1, characterized in that: The equivalent mechanical impedance of the vibration response region is calculated as follows: at the frequency fi, the ratio of the amplitude of the applied periodic pulse force to the amplitude of the vibration acceleration response measured in the vibration-dominant region Ai is taken as the amplitude of the equivalent mechanical impedance at the frequency point; the difference between the phase of the periodic pulse force and the phase of the vibration acceleration response is taken as the phase of the equivalent mechanical impedance.

9. The loss optimization method for power transformers according to claim 1, characterized in that: During the final tightening and fixing process, the attenuation of the static preload Fi is monitored in real time. If the attenuation of any static preload Fi exceeds 10% of its set value, the tightening process is paused, and the corresponding vibrator locking status and nearby fasteners are checked and adjusted.

10. The loss optimization method for power transformers according to claim 1, characterized in that: After removing all exciters, a scanning AC current excitation signal is applied to the transformer windings again to verify the optimized and tightened transformer body and check whether the amplitude of the original resonance peak to be optimized in its vibration spectrum has dropped below the preset qualified threshold.

Images