A method and system for loss optimization of a solid state transformer litz wire winding
By optimizing the Litz wire structure of the solid-state transformer using Fourier transform and genetic algorithm, a current distribution model was constructed, which solved the problems of concentrated loss and local overheating under non-sinusoidal excitation, achieving loss minimization and current balance across the entire frequency band, and improving equipment efficiency and lifespan.
Patent Information
- Application Number
- CN202610738806.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-27
- Publication Date
- 2026-07-14
AI Technical Summary
Under non-sinusoidal excitation, existing solid-state transformers exhibit concentrated losses and localized overheating in the Litz wire windings, and traditional design methods have failed to effectively optimize losses and current distribution across the entire frequency band.
The non-sinusoidal excitation signal is decomposed by Fourier transform, the dominant high and low frequencies are extracted, the structural parameters of the Litz line are optimized by genetic algorithm, a current distribution model is constructed, and the equivalent capacitive reactance parameter is fine-tuned to achieve loss balance.
It significantly reduces the losses of solid-state transformers across the entire frequency band, extends equipment lifespan, and improves energy transmission efficiency and reliability.
Smart Images

Figure CN122389380A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of transformer technology, and in particular to a method and system for optimizing the loss of Litz wire windings in solid-state transformers. Background Technology
[0002] Solid-state transformers (SSTs), as a new generation of power conversion equipment, integrate high-frequency transformers and power electronic converters, and have broad application prospects in fields such as new energy grid connection, smart grids, and electric vehicle charging. However, SSTs face a prominent problem in actual operation: their input / output signals often exhibit non-sinusoidal waveforms, containing abundant fundamental, low-order harmonics, and high-frequency harmonic components near the switching frequency. These multi-frequency excitation signals cause severe skin and proximity effects within the transformer windings, leading to a sharp increase in AC resistance, which in turn causes concentrated winding losses (copper losses), localized overheating, and a decrease in overall efficiency.
[0003] Litz wire, composed of multiple strands of insulated fine wires twisted together in a specific manner, is a common conductor form for high-frequency transformer windings, effectively suppressing the skin effect and proximity effect. However, traditional Litz wire design methods typically optimize parameters based on a single sinusoidal frequency (such as the switching frequency), neglecting the distribution characteristics of different frequency components in actual non-sinusoidal excitation signals. When SSTs operate under complex conditions such as wind power and photovoltaics, the amplitude and phase distribution of high-frequency harmonics dynamically change. Litz wires designed based on a single frequency struggle to achieve optimal loss suppression across the entire frequency band, resulting in high winding losses at some frequency components and uneven current distribution between winding layers, further exacerbating the risk of local overheating. While some existing technologies attempt to reduce transformer losses from the perspective of impedance matching or magnetic circuit optimization, few solutions combine the frequency domain distribution characteristics of non-sinusoidal excitation signals with Litz wire structural parameters (strand diameter, number of strands, strand pitch, winding arrangement) for joint optimization.
[0004] Therefore, how to systematically optimize the Litz wire winding structure based on the frequency distribution characteristics of actual signals, so as to minimize losses across the entire frequency band and balance the interlayer current distribution, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention
[0005] To address the aforementioned technical issues, this application provides a method and system for optimizing the loss of Litz wire windings in solid-state transformers. This method improves the energy transmission efficiency of solid-state transformers in complex non-sinusoidal excitation scenarios such as new energy power generation and industrial automation, thereby extending the service life of the equipment.
[0006] In a first aspect, this application provides a method for optimizing the loss of Litz wire windings in a solid-state transformer, the method comprising: Step S1: Collect the non-sinusoidal excitation signal during the operation of the solid-state transformer, perform Fourier transform decomposition on the non-sinusoidal excitation signal to obtain the corresponding frequency spectrum, extract the dominant frequency range of the high-frequency signal and the low-frequency signal based on the frequency spectrum, and determine the distribution characteristic data. Step S2: Using the distribution characteristic data as input, a genetic algorithm is used to simulate the loss response of the Litz wire winding under different structural parameters to obtain a set of Litz wire structural parameters that initially minimize the total loss. Step S3: Based on the structural parameters of the Litz wire, simulate the flow path of the current component between the winding layers of the Litz wire through electromagnetic field simulation, and construct a current distribution model between the winding layers; Step S4: Extract the loss density of each layer from the current distribution model, analyze the loss concentration area based on the loss density, and redistribute the current components until the loss distribution of each layer meets the equilibrium condition, and determine the final equilibrium distribution scheme. Step S5: Verify energy transfer using the balanced distribution scheme, calculate the overall efficiency and temperature rise, and output the verified Litz wire structure parameters and winding layout as the final optimization result.
[0007] Secondly, this application provides a system for optimizing the loss of Litz wire windings in a solid-state transformer, the system comprising: The acquisition module is used to acquire non-sinusoidal excitation signals during the operation of the solid-state transformer, perform Fourier transform decomposition on the non-sinusoidal excitation signals to obtain the corresponding frequency spectrum, extract the dominant frequency ranges of high-frequency and low-frequency signals based on the frequency spectrum, and determine the distribution characteristic data. The optimization module is used to take the distribution characteristic data as input, use a genetic algorithm to simulate the loss response of the Litz wire winding under different structural parameters, and obtain a set of Litz wire structural parameters that initially minimize the total loss. The module is used to simulate the flow path of current components between the winding layers of Litz wire through electromagnetic field simulation based on the structural parameters of Litz wire, and to build a current distribution model between the winding layers. The determination module is used to extract the loss density of each layer from the current distribution model, analyze the loss concentration area based on the loss density, and redistribute the current components until the loss distribution of each layer meets the equilibrium condition, and determine the final equilibrium distribution scheme. The verification module is used to verify energy transmission using the balanced distribution scheme, calculate the overall efficiency and temperature rise, and output the verified Litz wire structural parameters and winding layout as the final optimization result.
[0008] Compared with the prior art, the beneficial effects of the present invention are at least as follows: This application extracts low-frequency phase distribution and high-frequency amplitude distribution characteristic data by performing Fourier transform decomposition on the non-sinusoidal excitation signal. This allows subsequent optimization to accurately match the loss contribution of different frequency components under actual operating conditions, avoiding the parameter mismatch problem caused by traditional methods that only design Litz wire structures for a single frequency. A genetic algorithm is used with single-strand wire diameter, number of strands, pitch, and winding layer arrangement as optimization variables to construct a frequency-loss model and iteratively match it. This model can quickly converge to the structural parameter combination that minimizes total loss globally, effectively suppressing high-frequency additional losses caused by the skin effect and proximity effect. Furthermore, a frequency-dependent current distribution model is constructed through electromagnetic field simulation, extracting the main flow paths and verifying their stability. The equivalent capacitive reactance parameter is fine-tuned for loss-concentrated areas on the main flow paths, achieving active balanced current distribution among winding layers and significantly reducing the risk of local overheating. Finally, multi-condition energy transfer simulations verify the overall efficiency and temperature rise, ensuring the reliability of the optimized scheme under actual operating conditions. In summary, this application systematically improves the energy transmission efficiency of solid-state transformers in complex non-sinusoidal excitation scenarios such as new energy power generation and industrial automation, and extends the service life of equipment, demonstrating outstanding substantive features and significant progress. Attached Figure Description
[0009] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0010] Figure 1 This is a flowchart of a method for optimizing the loss of Litz wire windings in a solid-state transformer, as described in an embodiment of this application. Figure 2 The genetic algorithm iteration curve is shown in the embodiment of this application. Figure 3 This is a schematic diagram of a solid-state transformer Litz wire winding loss optimization system according to an embodiment of this application. Detailed Implementation
[0011] This application provides a method and system for optimizing the loss of Litz wire windings in solid-state transformers. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application 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 the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0012] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 The method for optimizing the loss of Litz wire windings in a solid-state transformer, as described in this application embodiment, includes: Step S1: Collect the non-sinusoidal excitation signal during the operation of the solid-state transformer, perform Fourier transform decomposition on the non-sinusoidal excitation signal to obtain the corresponding frequency spectrum, extract the dominant frequency range of the high-frequency signal and the low-frequency signal based on the frequency spectrum, and determine the distribution characteristic data.
[0013] In step S1, the Fourier transform decomposition of the non-sinusoidal excitation signal to obtain the corresponding frequency spectrum includes: collecting the non-sinusoidal excitation signal during the operation of the solid-state transformer, performing Fourier transform processing on the non-sinusoidal excitation signal, decomposing each frequency component and recording the corresponding frequency, amplitude and phase information; arranging the decomposed frequency components according to their frequency magnitude to construct a frequency spectrum with frequency on the horizontal axis and amplitude on the vertical axis.
[0014] Specifically, during the operation of a solid-state transformer, due to load fluctuations, high-frequency switching of power switching devices, and harmonic interference from the external power grid, the voltage or current signals acquired at its input terminal typically exhibit a non-sinusoidal waveform. This signal contains multiple frequency components, ranging from the low-frequency fundamental wave to high-frequency switching harmonics. Different frequency components have significantly different degrees of influence on the skin effect and proximity effect of the Litz wire winding. Therefore, it is first necessary to acquire these non-sinusoidal excitation signals using signal acquisition equipment. Specifically, high-precision voltage or current sensors can be installed at the input port or key nodes of the solid-state transformer. The sampling frequency of the sensors needs to be high enough to effectively capture high-frequency components; for example, it can be set to tens of thousands of samples per second. The acquisition should cover various operating states of the solid-state transformer, such as the startup phase, full-load operation, and low-load operation, to ensure the comprehensiveness and representativeness of the signal data. The acquired signals are stored in a data logger in time-series format.
[0015] After obtaining the time-series signal, it needs to be converted from the time domain to the frequency domain to analyze the specific impact of each frequency component on the loss of the Litz wire winding. Since the non-sinusoidal excitation signal of the solid-state transformer simultaneously contains a low-frequency fundamental component and high-frequency harmonic components generated by the operation of switching devices, and the skin effect and proximity effect vary significantly at different frequencies, it is essential to accurately determine the amplitude and phase of each frequency component. This step uses Fourier transform to decompose the signal. Fourier transform can decompose a complex non-sinusoidal time-domain waveform into a superposition of sinusoidal components of different frequencies. Specifically, a digital signal processor performs piecewise Fourier transform on the acquired signal data, outputting the frequency value, amplitude, and phase angle corresponding to each frequency component. The amplitude determines the weight of the frequency component in generating AC losses in the Litz wire winding, while the phase affects the peak current distribution after the superposition of the frequency components. These frequency, amplitude, and phase information together constitute the frequency domain representation of the signal, providing fundamental data for subsequent construction of the frequency spectrum and extraction of the dominant high- and low-frequency ranges.
[0016] Based on the frequency components obtained from the decomposition, they need to be arranged in order of frequency magnitude to construct a frequency spectrum that can intuitively reflect the signal energy distribution. Specifically, the amplitude corresponding to each frequency component is used as the vertical axis, and the frequency is used as the horizontal axis to plot a continuous distribution from low frequency to high frequency, thus obtaining the frequency spectrum of the non-sinusoidal excitation signal. This frequency spectrum clearly shows the weight of different frequency components in the signal. The low-frequency part usually corresponds to the fundamental frequency of the power grid and its lower harmonics, while the high-frequency part mainly comes from the operating frequency of the switching devices and its harmonic harmonics. By observing the frequency spectrum, it is possible to preliminarily identify which frequency components have significantly higher amplitudes than other components. These dominant frequency components will directly determine the magnitude of the AC loss of the Litz wire winding.
[0017] By accurately acquiring non-sinusoidal excitation signals and constructing frequency spectra, complex time-domain waveforms are transformed into clear frequency-domain distributions, enabling the quantification of the impact of each frequency component on the loss of the Litz line and providing an accurate data foundation for subsequent targeted optimization.
[0018] In step S1, determining the distribution characteristic data includes: separating the frequency spectrum into a low-frequency signal range and a high-frequency signal range according to a preset frequency boundary point; extracting the dominant frequency points and phase values with amplitudes higher than a first threshold in the low-frequency signal range, and the dominant frequency points and amplitudes with amplitudes higher than a second threshold in the high-frequency signal range; and integrating all the extracted dominant frequency points and their corresponding amplitudes and phases in frequency order into distribution characteristic data containing low-frequency phase distribution and high-frequency amplitude distribution.
[0019] Specifically, after constructing the frequency spectrum, to further extract the frequency components that have a dominant influence on the winding loss of the Litz wire from the complex frequency domain data, it is necessary to separate and process the high-frequency and low-frequency signals to quantify their respective contributions to the skin effect, proximity effect, and interlayer current phase distribution, thereby forming structured distribution characteristic data. Specifically, a frequency boundary point is pre-set based on the operating characteristics of the solid-state transformer. This boundary point can be determined according to the switching frequency of the power switching devices. For example, the frequency band below the switching frequency is classified as the low-frequency signal range, and the frequency band above the switching frequency is classified as the high-frequency signal range. The reason for this is that the low-frequency components mainly affect the DC resistance loss of the winding and the phase superposition effect of the interlayer current, while the high-frequency components directly determine the severity of the skin effect and proximity effect. Since the two have different focuses on optimizing the structural parameters of the Litz wire, they need to be processed separately. After separating the frequency spectrum according to this boundary point, frequency component data within the low-frequency signal range and the high-frequency signal range are obtained respectively.
[0020] Then, a first amplitude threshold is set in the low-frequency signal range to filter out dominant frequency points with amplitudes higher than this threshold, and the phase value corresponding to each dominant frequency point is recorded. A second amplitude threshold is set in the high-frequency signal range to filter out dominant frequency points with amplitudes higher than this threshold, and the amplitude corresponding to each dominant frequency point is recorded. The amplitude of high-frequency components directly reflects the energy intensity of high-frequency harmonics. The larger the amplitude of a high-frequency component, the more severe the skin effect and proximity effect caused by it; therefore, its amplitude needs to be recorded. The first and second thresholds can be set according to the statistical characteristics of the actual signal, for example, by taking a certain multiple of the average amplitude within their respective frequency ranges, to ensure that the extracted dominant frequency points can represent the main energy distribution of the signal. Finally, all the extracted dominant frequency points and their corresponding amplitudes and phases are integrated in ascending order of frequency to form a complete distribution characteristic data set. In this distribution characteristic data set, the low-frequency part is mainly based on phase distribution information, and the high-frequency part is mainly based on amplitude distribution information, comprehensively depicting the key influencing factors of non-sinusoidal excitation signals on the loss of Litz wire windings.
[0021] Through the above processing, this application can accurately extract the high-frequency amplitude distribution and low-frequency phase distribution that are most decisive for the loss of Litz wire winding from the original non-sinusoidal signal. This allows the subsequent genetic algorithm to optimize the structural parameters of Litz wire such as single strand diameter, number of strands, and pitch for different frequency bands. This effectively avoids the problem of the optimization target being out of focus due to ignoring frequency characteristics in traditional methods, and significantly improves the targeting and computational efficiency of loss optimization.
[0022] Step S2: Using the distributed characteristic data as input, a genetic algorithm is used to simulate the loss response of the Litz wire winding under different structural parameters to obtain the set of Litz wire structural parameters that initially minimize the total loss.
[0023] The step S2, obtaining the set of Litz wire structural parameters that initially minimizes the total loss, includes: simulating the distribution characteristic data using a genetic algorithm; constructing a frequency-loss model of the Litz wire winding using the single-strand diameter, number of strands, pitch, and winding layer arrangement as optimization variables; calculating the loss response value at each frequency component using the genetic algorithm, and performing a weighted summation based on the amplitude weights in the distribution characteristic data to obtain the total loss value; and generating a set of Litz wire structural parameters containing multiple sets of structural parameters based on the total loss value.
[0024] Specifically, after obtaining the distribution characteristic data, it needs to be transformed into an objective function and constraints that can be processed by a genetic algorithm, so as to search for the optimal combination that minimizes the total loss in the solution space of the Litz wire structural parameters. The distribution characteristic data contains the frequency value, amplitude, and phase information of each dominant frequency point in the non-sinusoidal excitation signal. The amplitude reflects the contribution of each frequency component to the total loss, while the phase provides a reference for subsequent current distribution simulation. In order to enable this frequency domain information to effectively guide the optimization of the Litz wire structural parameters, this step uses a genetic algorithm as the optimization tool. The frequency components and their amplitude weights in the distribution characteristic data are used as the basis for loss calculation, and the single strand diameter, number of strands, pitch, and winding layer arrangement of the Litz wire are used as optimization variables. The optimal solution is automatically found by simulating a biological evolution process.
[0025] First, the amplitude of each dominant frequency component in the distribution characteristic data is normalized by dividing the amplitude of each frequency component by the sum of the amplitudes of all dominant frequency components, resulting in a weight coefficient for each frequency component, ensuring that the sum of all weight coefficients is 1. This eliminates the difference in absolute amplitude between different signals, allowing the genetic algorithm to focus on the relative contribution of each frequency component and avoiding bias in the weighting result due to amplitude dimensions. The normalized weight coefficients will be used for the subsequent weighted summation of total losses. Next, a frequency-loss model of the Litz wire winding is constructed. This model is essentially an AC resistance calculation model. Its inputs are a set of Litz wire structural parameters (single strand diameter, number of strands, pitch, winding layer arrangement) and a frequency component, and the output is the winding loss at that frequency component. The core calculation logic of the model is as follows: For a given frequency component, the skin depth is calculated according to the skin depth formula. The skin depth reflects the degree to which high-frequency current concentrates on the surface of the conductor cross-section; the higher the frequency, the smaller the skin depth. Then, the radius of the single strand of the Litz wire is compared with the calculated... The skin depth is compared to the calculated value to obtain a first ratio. The skin effect factor is determined by this ratio using a standard engineering formula: when the first ratio is ≤1, it indicates that the single-strand wire diameter is sufficiently thin, and the skin effect can be ignored; in this case, the skin effect factor is 1. When the first ratio is >1, it indicates that the radius of the single-strand wire exceeds the skin depth, and the current is confined to a thin layer on the surface of the wire, reducing the effective conductive cross-sectional area and increasing the AC resistance. In this case, the skin effect factor is obtained by adding two parts: the first part is the ratio of the radius to twice the skin depth, and the second part is the fourth power of the ratio divided by three. Multiplying the calculated skin effect factor by the DC resistance of the single-strand wire yields the AC resistance of the single-strand wire at that frequency component. Through the above calculations, the frequency-loss model can accurately quantify the additional losses caused by the skin effect for each frequency component.
[0026] Subsequently, the proximity effect factors between layers and between strands are calculated based on the number of strands, the pitch, and the winding layer arrangement. Specifically, the proximity effect is an additional loss phenomenon caused by eddy currents induced in adjacent conductors by the alternating magnetic field generated by adjacent current-carrying conductors. This leads to a more uneven distribution of current across the conductor cross-section, thereby further increasing the AC resistance. For Litz wire windings, the proximity effect can be divided into two levels: first, the proximity effect between different strands within the same Litz wire; and second, the proximity effect between different winding layers. The strength of the proximity effect between strands mainly depends on the number of strands, the pitch, and the tightness of the strand arrangement across the cross-section. When the number of strands increases, the magnetic field coupling between the strands strengthens, and the induced eddy currents are larger; when the pitch decreases, the relative position change between the strands accelerates, the magnetic field linkage becomes more complex, and the proximity effect is correspondingly strengthened. In the calculation, firstly, based on the skin depth at a given frequency component, combined with the center distance between the strands and the strand pitch, an approximate value for the proximity effect factor is obtained through the classical proximity effect coefficient curve or empirical formula in electromagnetic field theory. For example, when the strand pitch is much larger than the strand diameter, the inter-strand coupling is weak, and the factor is close to 1; when the strand pitch and strand diameter are on the same order of magnitude or smaller, the factor will be significantly greater than 1. The interlayer proximity effect depends on the arrangement of the winding layers, including the interlayer spacing, interlayer insulation thickness, and the current direction (same or opposite) of each layer. For multi-layer windings, a particular layer is affected not only by the proximity effect between its own strands but also by the magnetic field generated by the current in adjacent layers. When calculating the interlayer proximity effect factor, it is usually based on a mirror current model, estimated according to the ratio of interlayer spacing to conductor diameter and the number of layers. The more layers and the smaller the interlayer spacing, the larger the interlayer proximity effect factor. Multiplying the skin effect factor, the inter-strand proximity effect factor, and the interlayer proximity effect factor together, and then multiplying by the DC resistance of the Litz wire at that frequency component, yields the total AC resistance corresponding to that frequency component. Multiplying this total AC resistance by the square of the current amplitude at that frequency component gives the winding loss at that frequency component.
[0027] After calculating the skin effect factor, interstrand proximity effect factor, and interlayer proximity effect factor, these three factors are multiplied to obtain the total AC resistance correction coefficient for that frequency component. Multiplying this correction coefficient by the DC resistance of the Litz wire for that frequency component yields the corresponding AC resistance value. The DC resistance can be calculated based on the total cross-sectional area of the Litz wire, the resistivity of the conductor material, and the total winding length. Multiplying the calculated AC resistance by the square of the current amplitude for that frequency component gives the winding loss value for that frequency component. Since the distributed characteristic data records multiple dominant frequency components, the frequency-loss model needs to perform the above calculations separately for each dominant frequency component.
[0028] When the genetic algorithm starts running, it first generates an initial population containing multiple sets of Litz wire structural parameters, each set of parameters being called an individual. For each individual, it is substituted into the frequency-loss model, and the loss value corresponding to each dominant frequency component in the distribution characteristic data is calculated sequentially. Then, these loss values are weighted and summed using the weighting coefficients obtained from the previous normalization to obtain the total loss value of that individual. The physical meaning of the weighted summation is that the frequency component with the larger amplitude contributes more to the actual total winding loss, so it should be given a higher weight in the optimization, so that the genetic algorithm prioritizes reducing the loss under these dominant frequency components. The total loss value is the fitness of that individual; the smaller the total loss, the higher the fitness. The genetic algorithm performs selection operations based on fitness, retaining individuals with higher fitness, and then generates a new generation of population through crossover (exchanging some structural parameters between two excellent individuals) and mutation (randomly fine-tuning the single-strand diameter or pitch of an individual). Repeat the above evaluation, selection, crossover, and mutation process until the total loss value of individuals in the population no longer decreases significantly, i.e., converges to the minimum value. At this point, the corresponding set of Lids line structural parameters is the set of structural parameters that initially minimizes the total loss.
[0029] like Figure 2 The figure shows the iterative curve of the genetic algorithm. In this embodiment, a non-sinusoidal excitation signal actually collected from a solid-state transformer is used as input. The genetic algorithm is run to optimize the structural parameters of the Litz line, and the optimal individual and the population average total loss value in each generation are recorded. In the initial population (generation 0), the optimal individual total loss is 62.0W, and the population average total loss is 69.0W. After a rapid decrease in the first 20 generations, the optimal individual loss drops to 31.0W, and the average loss drops to 39.0W. After 50 generations, the rate of decrease slows down, and by generation 100, the optimal individual loss converges to 12.5W, and the population average loss converges to 15W. Throughout the iteration process, the optimal individual loss is always lower than the population average, and both eventually stabilize within a small fluctuation range, indicating that the genetic algorithm can quickly and stably converge to the global optimum, verifying the efficiency and reliability of the proposed method in optimizing the structural parameters of the Litz line.
[0030] Through the above process, the abstract frequency domain distribution characteristics are transformed into specific structural parameter optimization targets, avoiding the parameter mismatch problem caused by only considering a single frequency or ignoring the amplitude weight in the traditional Litz line design, and providing an accurate starting point for structural parameters for subsequent iterative matching and current distribution optimization.
[0031] Step S3: Based on the structural parameters of the Litz wire, simulate the flow path of the current component between the winding layers of the Litz wire through electromagnetic field simulation, and construct a current distribution model between the winding layers.
[0032] Step S3 includes: importing the structural parameters of the Litz wire into the electromagnetic field simulation model; setting the frequency components, amplitude, and phase of the non-sinusoidal excitation signal in the simulation model; calculating the flow path of the current between the winding layers of the Litz wire at each frequency; extracting the current density distribution on each winding layer based on the simulation results; analyzing the interlayer distribution differences between the high-frequency current components and the low-frequency current components; and constructing a frequency-dependent current distribution model containing the interlayer current density matrix based on the distribution differences.
[0033] Specifically, after obtaining the structural parameters of the Litz wire, in order to further verify the current distribution characteristics of these parameters under the action of an actual electromagnetic field, it is necessary to simulate the actual flow path of the current between the winding layers through electromagnetic field simulation, thereby constructing a current distribution model that can reflect the frequency-dependent characteristics, and providing a basis for the subsequent analysis of the loss concentration area.
[0034] Specifically, the Litz wire structure parameters obtained through iterative matching optimization are imported into an electromagnetic field simulation model. This electromagnetic field simulation model is a numerical calculation tool based on the finite element method, used to solve Maxwell's equations for the electromagnetic field distribution under given conductor structure and excitation conditions. In this application, the model input includes: the three-dimensional geometry of the Litz wire winding (size, spacing, and arrangement order of each layer of winding), material properties (conductivity of copper, dielectric constant of insulation), and excitation source settings. The excitation source settings are a crucial step, requiring the loading of each frequency component of the non-sinusoidal excitation signal obtained in step S1, along with its corresponding amplitude and phase, into the simulation model one by one. That is, sinusoidal excitation is applied to each dominant frequency point, and the current distribution at that frequency is recorded. The reason for simulating each frequency component separately is that the skin effect and proximity effect vary significantly at different frequencies, and superposition simulation cannot distinguish the contribution of each frequency. For each frequency component, the simulation model calculates the current density distribution on the cross-section of each layer of Litz wire winding by solving the magnetic field equation, thereby obtaining the current flow path from the input end to the output end between layers. At high frequencies, the current tends to flow along the outer winding because the leakage flux and inductive reactance of the outer winding are smaller; at low frequencies, the current is more evenly distributed across the winding layers.
[0035] After completing the simulation of all frequency components, the current density values of each winding layer are extracted based on the simulation results and organized into a current density matrix according to frequency and layer position. Then, the distribution differences between high-frequency and low-frequency current components in each layer are compared. For example, whether the current density of high-frequency components in the outer winding is significantly higher than that in the inner winding, and whether the low-frequency components show a uniform distribution, etc. Based on these distribution differences, a frequency-dependent current distribution model is constructed. This model uses frequency as the independent variable and the current density distribution ratio of each winding layer as the dependent variable. It can be stored in matrix form, where the row index represents the frequency component, the column index represents the winding layer number, and the matrix element is the current density value of the winding layer at the corresponding frequency. This model can clearly describe the distribution law of different frequency components among winding layers, providing direct numerical basis for the subsequent location of loss concentration areas and fine-tuning of equivalent capacitive reactance parameters.
[0036] Through the above electromagnetic field simulation and current distribution model construction, this application closely integrates the structural parameters of the Litz wire with the frequency domain characteristics of the actual non-sinusoidal excitation signal, realizing high-precision quantification of the interlayer current distribution of the winding. This model not only reveals the concentration trend of high-frequency current in the outer winding, but also quantifies the specific values of the current density of each layer, providing a clear direction for subsequent identification of loss concentration areas and targeted adjustment of the equivalent capacitive reactance parameters, thereby effectively avoiding the problems of local overheating and efficiency reduction caused by uneven current distribution.
[0037] Step S3 and subsequent steps include: inputting test signals of different frequencies into the current distribution model to verify the stability of the current distribution at different frequencies; extracting the flow path corresponding to the interlayer region of the winding where the current density exceeds a preset threshold as the main flow path, and integrating the topology information of the main flow path into the current distribution model.
[0038] Specifically, after constructing the frequency-dependent current distribution model, to further verify its reliability and adaptability under different excitation conditions and extract the critical path information most instructive for loss analysis, it is necessary to verify the model by inputting multi-frequency test signals and selecting the main flow paths with high current density based on preset thresholds. Specifically, for the constructed current distribution model, test signals of different frequencies are input. These test signals can be the dominant frequency components already present in step S1, or several verification frequency points selected near the dominant frequency, such as several interpolation frequencies selected in both the low-frequency and high-frequency bands. After inputting the test signals, the model outputs the current density distribution of each winding layer at the test frequency through interpolation or extrapolation based on its internally stored interlayer current density matrix. The output results at different test frequencies are compared to determine whether the current distribution exhibits a monotonous and smooth trend with frequency change, and whether there are any abnormal jumps or concentrations. If, at a certain test frequency, the current density of a certain winding layer shows a significant jump compared to adjacent frequencies, it indicates that the model's stability near that frequency is insufficient, and it is necessary to return to step S3 to readjust the simulation parameters or refine the frequency sampling points. After the stability verification is passed, it is considered that the current distribution model can reliably reflect the distribution law of different frequency components between winding layers under actual operating conditions.
[0039] Subsequently, the main flow paths are extracted from the current distribution model that has passed stability verification. Specifically, for each frequency component, all current flow channels between winding layers are traversed, and flow paths corresponding to regions where the current density exceeds a preset threshold are marked as candidate paths. The preset threshold can be determined based on the statistical distribution of current density in all inter-layer regions; for example, two or three times the average current density can be used as the threshold value. Regions exceeding this threshold value indicate highly concentrated current, which is the main channel leading to increased local losses. All inter-layer regions refer to all two-dimensional interface regions between any two adjacent physical layers in the winding, excluding inter-strand gaps within the same layer. These candidate paths are categorized by frequency component, and the winding layer number, peak current density, and corresponding frequency range traversed by each path are recorded to form the topological information of the main flow paths. This topological information includes the path's starting point, ending point, the inter-layer interfaces it passes through, and the current density distribution on each segment. The topological information of the main flow paths is integrated into the current distribution model, that is, a path topology layer is added to the original frequency-layer current density matrix, enabling the model to not only describe the current density of each winding layer but also trace the specific channels through which current transfers between layers. The integrated model can directly identify which interlayer regions are the main channels of high-frequency current, thus providing accurate spatial positioning information for targeted adjustment of equivalent capacitive reactance parameters and guidance of current redistribution in subsequent step S4.
[0040] Through the above stability verification and main flow path extraction, this application further enhances the engineering practical value of the current distribution model, ensuring that the loss balancing adjustment can be accurately applied to the critical path that truly leads to loss concentration, and avoiding efficiency loss caused by blind adjustment.
[0041] Step S4: Extract the loss density of each layer from the current distribution model, analyze the loss concentration area based on the loss density, and redistribute the current components until the loss distribution of each layer meets the equilibrium condition, and determine the final equilibrium distribution scheme.
[0042] Step S4 includes: extracting the current density and flow path information of each winding layer from the current distribution model, and calculating the loss density of each layer; marking the winding layer regions with loss densities higher than the average loss density by a preset multiple as loss concentration regions; for the loss concentration regions traversed by the main flow path, changing the current distribution ratio on the main flow path by fine-tuning the equivalent capacitive reactance parameters of the corresponding winding layers, recalculating the loss density of each layer based on the adjusted equivalent capacitive reactance parameters, and determining whether the standard deviation between the loss densities of each layer is less than a preset threshold or whether the ratio of the highest loss to the lowest loss is less than a preset ratio; if not, continuing to fine-tune the equivalent capacitive reactance parameters and recalculating until the equilibrium condition is met; integrating the loss density, current distribution, and corresponding equivalent capacitive reactance parameters of each layer when the equilibrium condition is met into the final equilibrium distribution scheme.
[0043] Specifically, after completing the construction of the current distribution model and extracting the main flow paths, in order to reduce local overheating and energy loss caused by uneven current distribution, it is necessary to identify the loss concentration area based on the loss density distribution in the model, and guide the current redistribution by fine-tuning the capacitive reactance parameter until the loss of each layer reaches a balanced state, thereby forming the final balanced distribution scheme.
[0044] Specifically, the current density and flow path information for each winding layer are extracted from the current distribution model. Current density reflects the amount of current passing through a unit cross-sectional area and is the basis for calculating losses; the flow path indicates the specific channels through which current transfers between layers. Based on the extracted current density and the resistivity of the Litz wire winding, the loss density of each winding layer is calculated, i.e., the power loss per unit volume or unit area. Loss density directly reflects the heat generation intensity of that winding layer during operation and is a key indicator for determining the risk of overheating. The loss density of each layer is statistically analyzed, and its average value is calculated. Winding layer regions with loss densities exceeding a preset multiple (e.g., 1.5 or 2 times) of the average loss density are marked as loss concentration areas. These areas are typically where current is highly concentrated and are the main causes of overall efficiency degradation and excessive local temperature rise.
[0045] The primary flow path is the channel that carries the highest current density and contributes the most losses. In the current distribution model, there are multiple current flow paths between different winding layers, but only a few of these paths have current densities significantly higher than the average. These paths are the root cause of localized overheating and overall efficiency degradation. Therefore, it is necessary to adjust the equivalent capacitive reactance parameter of the corresponding winding layer in the loss-concentrated area traversed by the primary flow path to change the current distribution. The equivalent capacitive reactance parameter is provided by adjustable capacitors or reactors connected in parallel or series between winding layers. The capacitance or inductance value is changed in real time by the control circuit (e.g., using varactor diodes, switched capacitor arrays, or magnetically controlled reactors), thereby adjusting the shunting ratio of high-frequency current between layers without changing the physical structure of the transformer. Specifically, if the loss-concentrated area on a certain primary flow path exhibits excessively high current density, the equivalent capacitive reactance of the winding layer corresponding to that layer can be appropriately reduced (e.g., increasing the parallel capacitance value to reduce the capacitive reactance) or the equivalent capacitive reactance of adjacent layers can be increased, allowing more high-frequency current to flow through other paths, thereby reducing the current density on that primary flow path.
[0046] After adjusting the equivalent capacitive reactance parameters, the loss density of each layer is recalculated, and the degree of balance in the inter-layer loss distribution is evaluated. The balance condition can be quantified as follows: the standard deviation between the loss densities of each layer is less than a preset threshold, or the ratio of the highest loss density to the lowest loss density is less than a preset ratio (e.g., not exceeding 2:1). If the balance condition is not currently met, the equivalent capacitive reactance parameters of the corresponding winding layer are fine-tuned, and the loss density is recalculated until the balance condition is met. The step size of each fine-tuning can be dynamically set according to the current loss deviation to avoid over-adjustment leading to new imbalances. When the balance condition is met, the loss density, current distribution data of each layer, and the corresponding capacitive reactance parameters are integrated into the final balanced distribution scheme. This scheme records the loss density, current density, and equivalent capacitive reactance parameter values set to achieve this balance for each winding layer in the balanced state, and can be directly used for subsequent energy transmission verification and parameter configuration of the actual transformer.
[0047] Through the above-mentioned process of identifying concentrated loss areas and fine-tuning equivalent capacitive reactance parameters, this application can accurately locate the critical path leading to local overheating. By changing the interlayer capacitance characteristics to guide the current to be reasonably distributed, it effectively reduces the excessive concentration of high-frequency current in the outer winding, making the loss distribution of each layer more uniform, thereby significantly reducing the risk of local overheating and improving the overall operating efficiency and long-term reliability of the solid-state transformer.
[0048] Step S5: Verify energy transfer using a balanced distribution scheme, calculate the overall efficiency and temperature rise, and output the verified Litz wire structural parameters and winding layout as the final optimization result.
[0049] Step S5 includes: obtaining the Litz wire structural parameters, winding layout, and corresponding loss density and current distribution data of each layer in the balanced distribution scheme, and substituting them into the energy transmission simulation model to simulate the current and loss distribution of the solid-state transformer under various operating conditions, including full load, low load, and dynamic load, and calculating the overall energy transmission efficiency and the temperature rise value of each winding layer; if the energy transmission efficiency is higher than the preset efficiency threshold and the temperature rise value of each layer is lower than the preset temperature rise threshold, then the verification is confirmed to be successful; the verified Litz wire structural parameters and winding layout are output as the final optimization result.
[0050] Specifically, after determining the balanced distribution scheme, to verify its effectiveness and reliability under actual operating conditions, energy transfer simulation verification is required for the Litz wire structural parameters and winding layout. The overall efficiency and temperature rise of each winding layer are calculated to assess whether the design objectives are met, and the verified parameters are output as the final optimization results. Specifically, the balanced distribution scheme data integrated in step S4 is obtained. This data includes Litz wire structural parameters (single strand diameter, number of strands, pitch), winding layout (arrangement order of each layer of windings, layer spacing, insulation method), and corresponding loss density and current distribution information for each layer. This data is substituted into the energy transfer simulation model. The energy transfer simulation model uses the finite element method, establishing a three-dimensional electromagnetic-thermal coupling model of the solid-state transformer winding in commercial software (such as COMSOL Multiphysics). The electromagnetic field solution uses a frequency domain solver, calculating the current density distribution at each frequency component and superimposing it to obtain the total loss distribution. The thermal field solution uses a transient or steady-state solver. The boundary conditions are set as follows: ambient temperature 25℃, convective heat transfer coefficient between the winding surface and air 10 W / (m²·K), and the outer shell is set to natural convection or forced air cooling (wind speed determined according to actual design). The mesh generation must ensure at least 2-3 elements within the skin depth layer. During simulation, the rated voltage excitation is applied first, the loss distribution is calculated, and then the loss is added as a heat source to the thermal model. The solution is iteratively solved until temperature convergence. The above process establishes the simulation baseline under a single rated operating condition.
[0051] To comprehensively evaluate the adaptability of the balanced distribution scheme, it is necessary to simulate the performance of solid-state transformers under various operating conditions, including full-load operation (rated power output), low-load operation (e.g., 20% or 30% of rated power), and dynamic load changes (e.g., a sudden switch from light load to full load, or simulating power fluctuations in actual operating conditions). For each operating condition, the simulation model calculates the current distribution, loss distribution, and temperature rise of each winding layer inside the transformer by solving the electromagnetic field equation and the heat conduction equation. The overall energy transfer efficiency, equal to the ratio of output power to input power, reflects the effectiveness of the optimization scheme in suppressing energy conversion losses; the temperature rise of each winding layer is calculated based on loss density and heat dissipation conditions, and is a key indicator for assessing the risk of localized overheating.
[0052] The calculated overall efficiency is compared with a preset efficiency threshold (e.g., 95%), and the temperature rise of each layer is compared with a preset temperature rise threshold, such as a safety upper limit set according to the temperature resistance rating of the insulation material. If the overall efficiency is higher than the preset threshold, and the temperature rise of all winding layers is lower than the preset temperature rise threshold, then the balanced distribution scheme is confirmed to have passed verification. After successful verification, the corresponding Litz wire structural parameters and winding layout are output as the final optimization results for use in the actual design or production of solid-state transformers. If the verification fails, the fallback path is determined according to the following rules: If the overall efficiency is below the threshold but the temperature rise of each layer is within acceptable limits, it indicates that the total loss is too high. Return to step S2, rerun the genetic algorithm and iterative matching, and tighten the total loss convergence condition (e.g., reduce the convergence tolerance by half). If the temperature rise of a certain layer exceeds the limit and the loss density of that layer is significantly higher than the average (more than 1.5 times the average), it indicates that the inter-layer distribution is unbalanced. Return to step S4, increase the number of iterations for equivalent capacitive reactance adjustment, or adjust the step size coefficient. If both low efficiency and excessive temperature rise exist simultaneously, prioritize returning to step S4 for a maximum of 5 readjustments. If the conditions are still not met, return to step S2. To prevent infinite loops, the maximum number of fallbacks is set to 3. If the conditions are still not met after 3 iterations, output the current optimal result and issue a warning.
[0053] Through the above energy transmission verification, this application ensures that the optimized Litz wire winding not only meets the theoretical design goals in terms of minimizing losses across the entire frequency band and achieving balanced distribution of interlayer current, but also maintains efficient and stable energy transmission capabilities under actual multi-condition operation. This effectively avoids the risk of equipment failure due to local overheating or substandard efficiency, providing a solid technical guarantee for the reliable application of solid-state transformers in complex scenarios such as new energy power generation and industrial automation.
[0054] The above describes a method for optimizing the loss of a Litz wire winding in a solid-state transformer according to an embodiment of this application. The following describes a system for optimizing the loss of a Litz wire winding in a solid-state transformer according to an embodiment of this application. Please refer to [link / reference]. Figure 3The Litz wire winding loss optimization system for a solid-state transformer in this application embodiment includes: The acquisition module is used to acquire non-sinusoidal excitation signals during the operation of the solid-state transformer, perform Fourier transform decomposition on the non-sinusoidal excitation signals to obtain the corresponding frequency spectrum, extract the dominant frequency range of high-frequency and low-frequency signals based on the frequency spectrum, and determine the distribution characteristic data. The optimization module is used to simulate the loss response of the Litz wire winding under different structural parameters using a genetic algorithm, taking the distribution characteristic data as input, and to obtain the set of Litz wire structural parameters that initially minimize the total loss. The module is used to simulate the flow path of current components between the winding layers of Litz wire through electromagnetic field simulation based on the structural parameters of Litz wire, and to build a current distribution model between the winding layers. The determination module is used to extract the loss density of each layer from the current distribution model, analyze the loss concentration area based on the loss density, and redistribute the current components until the loss distribution of each layer meets the equilibrium condition, and determine the final equilibrium distribution scheme. The verification module is used to verify energy transfer using a balanced distribution scheme, calculate the overall efficiency and temperature rise, and output the verified Litz wire structural parameters and winding layout as the final optimization results.
[0055] In summary, this application addresses the problem of loss concentration and efficiency degradation in Litz wire windings caused by skin effect and proximity effect when solid-state transformers are subjected to non-sinusoidal excitation signals in scenarios such as new energy power generation and industrial automation. It proposes a complete technical solution from frequency domain analysis to structural optimization and electromagnetic verification. First, the non-sinusoidal signal is decomposed into a frequency spectrum using Fourier transform, extracting low-frequency phase distribution and high-frequency amplitude distribution characteristics. Then, a genetic algorithm combined with a frequency-loss model is used, with the amplitude weights of each frequency component as the optimization basis, to initially obtain the Litz wire structural parameters with the minimum total loss. Subsequently, a frequency-dependent current distribution model is constructed using electromagnetic field simulation, extracting the main flow paths and verifying their stability. For loss concentration areas on the main flow paths, the interlayer equivalent capacitive reactance parameter is fine-tuned to guide current redistribution, making the loss density of each layer tend to be balanced. Finally, multi-condition energy transfer simulation is used to verify the overall efficiency and temperature rise, ensuring that the optimized parameters meet engineering requirements. This solution systematically solves the key technical problems of loss suppression and current balance of Litz wire windings across the entire frequency band, significantly improving the operating efficiency and reliability of solid-state transformers.
[0056] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.
[0057] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0058] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A method for optimizing the loss of Litz wire windings in a solid-state transformer, characterized in that, The method includes: Step S1: Collect the non-sinusoidal excitation signal during the operation of the solid-state transformer, perform Fourier transform decomposition on the non-sinusoidal excitation signal to obtain the corresponding frequency spectrum, extract the dominant frequency range of the high-frequency signal and the low-frequency signal based on the frequency spectrum, and determine the distribution characteristic data. Step S2: Using the distribution characteristic data as input, a genetic algorithm is used to simulate the loss response of the Litz wire winding under different structural parameters to obtain a set of Litz wire structural parameters that initially minimize the total loss. Step S3: Based on the structural parameters of the Litz wire, simulate the flow path of the current component between the winding layers of the Litz wire through electromagnetic field simulation, and construct a current distribution model between the winding layers; Step S4: Extract the loss density of each layer from the current distribution model, analyze the loss concentration area based on the loss density, and redistribute the current components until the loss distribution of each layer meets the equilibrium condition, and determine the final equilibrium distribution scheme. Step S5: Verify energy transfer using the balanced distribution scheme, calculate the overall efficiency and temperature rise, and output the verified Litz wire structure parameters and winding layout as the final optimization result.
2. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, In step S1, the non-sinusoidal excitation signal is decomposed using a Fourier transform to obtain the corresponding frequency spectrum, including: Non-sinusoidal excitation signals are collected during the operation of a solid-state transformer. The non-sinusoidal excitation signals are processed by Fourier transform to decompose each frequency component and record the corresponding frequency, amplitude and phase information. The frequency components obtained by decomposition are arranged according to their frequency magnitude to construct a frequency spectrum with frequency on the horizontal axis and amplitude on the vertical axis.
3. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, Step S1 involves determining the distribution characteristic data, including: Based on the preset frequency boundary point, the frequency spectrum is separated into a low-frequency signal range and a high-frequency signal range. The dominant frequency points and phase values with amplitudes higher than a first threshold in the low-frequency signal range, and the dominant frequency points and amplitudes with amplitudes higher than a second threshold in the high-frequency signal range are extracted respectively. All the extracted dominant frequency points and their corresponding amplitudes and phases are integrated in frequency order into distribution characteristic data containing low-frequency phase distribution and high-frequency amplitude distribution.
4. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, Step S2 involves obtaining the set of Litz line structural parameters that initially minimize the total loss, including: A genetic algorithm is used to simulate the distribution characteristic data. Using the genetic algorithm, the single strand diameter, number of strands, pitch and winding layer arrangement in the Litz wire structural parameters are used as optimization variables to construct a frequency-loss model of the Litz wire winding. The loss response value at each frequency component is calculated using the genetic algorithm, and the total loss value is obtained by weighted summation based on the amplitude weights in the distribution characteristic data. Based on the total loss value, a set of Litz line structural parameters containing multiple sets of structural parameters is generated.
5. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, Step S3 includes: The structural parameters of the Litz wire are imported into an electromagnetic field simulation model. In the simulation model, the frequency components, amplitudes, and phases of the non-sinusoidal excitation signal are set, and the flow paths of the current between the winding layers of the Litz wire at each frequency are calculated. Based on the simulation results, the current density distribution on each winding layer is extracted, and the interlayer distribution differences between the high-frequency current components and the low-frequency current components are analyzed. Based on the distribution differences, a frequency-dependent current distribution model containing the interlayer current density matrix is constructed.
6. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, Step S3 is followed by: For the current distribution model, test signals of different frequencies are input to verify the stability of the current distribution at different frequencies; the flow path corresponding to the winding interlayer region where the current density exceeds a preset threshold is extracted as the main flow path, and the topology information of the main flow path is integrated into the current distribution model.
7. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, Step S4 includes: The current density and flow path information of each winding layer are extracted from the current distribution model, and the loss density of each layer is calculated. The winding layer regions with loss densities higher than the average loss density by a preset multiple are marked as loss concentration regions. For the loss concentration regions traversed by the main flow path, the equivalent capacitive reactance parameters of the corresponding winding layers are fine-tuned to change the current distribution ratio on the main flow path. Based on the adjusted equivalent capacitive reactance parameters, the loss density of each layer is recalculated, and it is determined whether the standard deviation between the loss densities of each layer is less than a preset threshold or whether the ratio of the highest loss to the lowest loss is less than a preset ratio. If not, the equivalent capacitive reactance parameters are fine-tuned and recalculated until the equilibrium condition is met. The loss density, current distribution, and corresponding equivalent capacitive reactance parameters of each layer when the equilibrium condition is met are integrated into the final equilibrium distribution scheme.
8. The method for optimizing the loss of Litz wire windings in a solid-state transformer according to claim 1, characterized in that, Step S5 includes: Obtain the Litz wire structural parameters, winding layout, and corresponding loss density and current distribution data of each layer in the balanced distribution scheme, and substitute them into the energy transmission simulation model to simulate the current and loss distribution of the solid-state transformer under various operating conditions, including full load, low load, and dynamic load. Calculate the overall energy transmission efficiency and the temperature rise value of each winding layer. If the energy transmission efficiency is higher than the preset efficiency threshold and the temperature rise value of each layer is lower than the preset temperature rise threshold, the verification is confirmed to be successful. The verified Litz wire structural parameters and winding layout are output as the final optimization result.
9. A system for optimizing the loss of Litz wire windings in a solid-state transformer, used to implement the method for optimizing the loss of Litz wire windings in a solid-state transformer as described in any one of claims 1-8, characterized in that, The system includes: The acquisition module is used to acquire non-sinusoidal excitation signals during the operation of the solid-state transformer, perform Fourier transform decomposition on the non-sinusoidal excitation signals to obtain the corresponding frequency spectrum, extract the dominant frequency ranges of high-frequency and low-frequency signals based on the frequency spectrum, and determine the distribution characteristic data. The optimization module is used to take the distribution characteristic data as input, use a genetic algorithm to simulate the loss response of the Litz wire winding under different structural parameters, and obtain a set of Litz wire structural parameters that initially minimize the total loss. The module is used to simulate the flow path of current components between the winding layers of Litz wire through electromagnetic field simulation based on the structural parameters of Litz wire, and to build a current distribution model between the winding layers. The determination module is used to extract the loss density of each layer from the current distribution model, analyze the loss concentration area based on the loss density, and redistribute the current components until the loss distribution of each layer meets the equilibrium condition, and determine the final equilibrium distribution scheme. The verification module is used to verify energy transmission using the balanced distribution scheme, calculate the overall efficiency and temperature rise, and output the verified Litz wire structural parameters and winding layout as the final optimization result.
