Method for predicting magnetic loss of magnetic element based on modified SteinMazz equation
By modifying the Steinmetz equation and introducing the temperature term, combined with the neural network fitting parameters, the problem of insufficient prediction accuracy of the Steinmetz equation in temperature-sensitive environments is solved, and higher-precision magnetic loss prediction is achieved.
Patent Information
- Application Number
- CN202510809283.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-17
- Publication Date
- 2025-09-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing Steinmetz equation has large errors when predicting the magnetic loss of magnetic components in temperature-sensitive environments and fails to effectively consider the influence of temperature factors.
Based on the modified Steinmetz equation, temperature-related terms are introduced, and the parameters k1, ɑ1, β1 and correction coefficient c are fitted by the neural network fitting method. Combined with the temperature T, the Steinmetz equation is modified to improve the prediction accuracy.
The prediction accuracy of magnetic loss is significantly improved under different temperature conditions, especially in high frequency and high power density scenarios, providing more accurate loss prediction results.
Smart Images
Figure CN120703653A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a method for predicting magnetic loss of a magnetic element. Background Art
[0002] Core loss is the energy loss in magnetic components (such as transformers and inductors) caused by magnetic domain motion and eddy current effects in an alternating magnetic field. It primarily includes hysteresis loss, eddy current loss, and residual loss. The Steinmetz equation (SE), proposed by German engineer Charles Steinmetz in 1892, is the most classic empirical model for describing core loss. It uses experimental data to fit the relationship between loss, frequency, and magnetic flux density, and remains a crucial tool for quickly estimating losses in engineering design.
[0003] The Steinmetz equation is used under sinusoidal excitation (excitation refers to the current passing through the coil of the magnetic component to generate a magnetic field). The core loss calculation formula is as follows:
[0004]
[0005] Where P is the core loss; f is the excitation frequency; B m is the peak value of magnetic flux density; k1, ɑ1, and β1 are all material characteristic fitting parameters, which are given by the core material manufacturer through standard testing. Generally, 1<ɑ1<3, 2<β1<3. In this model, the core loss P is defined as the product of frequency f and magnetic flux density B. m Based on this relationship, the Steinmetz equation can be derived and its analytical solution can be obtained, so that the corresponding magnetic characteristic curve and related parameters can be calculated.
[0006] Although the Steinmetz equation shows good adaptability under certain specific conditions, the model does not take the influence of temperature factors into account, so the error is relatively large in temperature-sensitive environments. Summary of the Invention
[0007] Purpose of the invention: In view of the above-mentioned existing technologies, a method for predicting the magnetic loss of magnetic components based on the modified Steinmetz equation is proposed, which is suitable for loss prediction under different temperature change conditions.
[0008] Technical solution: A method for predicting magnetic loss of magnetic components based on the modified Steinmetz equation. The modified Steinmetz equation is:
[0009]
[0010] Where T is the internal temperature of the magnetic component, c is the correction factor, P is the core loss, f is the excitation frequency, and B mis the peak value of magnetic flux density; k1, ɑ1, and β1 are all material characteristic fitting parameters; the parameters k1, ɑ1, and β1 are given by the magnetic component manufacturer, or the correction coefficient c is obtained by fitting experimental data;
[0011] When predicting magnetic loss, the temperature T is collected and the prediction result is output after calculation according to the modified Steinmetz equation.
[0012] Furthermore, when fitting parameters through experimental data, the experimental data are first preprocessed, including: cleaning the data, removing duplicate values and outliers, and ensuring the integrity and consistency of the data; normalizing the numerical data to eliminate the influence of different dimensions on the model; at the same time, encoding the categorical variables for application to the neural network model; then using the neural network fitting method and through nonlinear regression, the parameters required for fitting in the modified Steinmetz equation are obtained.
[0013] An electronic device comprises a memory, a processor and a computer program stored in the memory and runnable on the processor. When the processor executes the program, the magnetic loss prediction method of the magnetic element is implemented.
[0014] Beneficial effects: Based on the Steinmetz equation, the present invention proposes a power function temperature correction equation, which takes into account the influence of temperature on core loss by introducing temperature-related terms. By introducing the temperature term, the influence of temperature on core loss can be taken into account, and the core loss can be predicted more accurately under different temperature conditions. The modified Steinmetz equation proposed by the present invention has a significant improvement in prediction accuracy compared to the traditional Steinmetz equation, especially in the case of larger loss values. It can better capture the influence of external factors such as temperature on core loss, thereby providing more accurate prediction results. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] Figure 1 The process of neural network fitting is shown in the figure;
[0016] Figure 2 This is a comparison chart of the prediction accuracy between the modified Steinmetz equation proposed in the present invention and the traditional Steinmetz equation. DETAILED DESCRIPTION
[0017] The present invention will be further explained below with reference to the accompanying drawings.
[0018] First, the k1, ɑ1, and β1 parameters of the Steinmetz equation are fitted based on the experimental data.
[0019] Select typical magnetic components and ensure that the component structure and core material are clear. Use professional magnetic measurement equipment (such as BH analyzer, temperature control test box) to collect magnetic flux density, loss data and temperature data. The test temperature is 25℃~90℃, and the experiment is carried out step by step in 5℃ or 10℃. Ensure that the temperature is stable before measuring the data to ensure data accuracy. At each temperature point, measure the core loss under excitation conditions of different frequencies (such as from 10kHz to 500kHz) and record the peak value of the magnetic flux density. Repeat each measurement condition multiple times and take the average value to reduce random errors.
[0020] To fit the parameters of the Steinmetz equation from the experimental data, a series of preprocessing operations were first performed on the data. These included: cleaning the data to remove duplicates and outliers to ensure data integrity and consistency; normalizing the numerical data to eliminate the influence of different dimensions on the model; and encoding the categorical variables for better application in the neural network model.
[0021] On this basis, the present invention adopts a neural network fitting method, which is not only used to predict the predicted value within a certain period of time in the future, but also dynamically corrects the current time point based on the obtained predicted value, thereby further improving the prediction accuracy and reliability of the model. Neural network fitting technology is an efficient nonlinear data fitting method that relies on artificial neural networks to simulate and learn the relationship between data. During the training phase, its internal parameters (also called weights and biases) are adjusted in order to minimize the deviation between the predicted value and the actual value. The process of neural network fitting is as follows Figure 1 As shown. Through nonlinear regression, the fitting parameters of the traditional Steinmetz equation are: k1=1.4997, ɑ1=1.4296, β1=2.4713. These parameters show that the core loss increases with the frequency f 1.4385 The rate of increase is as the peak value of magnetic flux density increases. m 2.4324 growth rate.
[0022] The loss characteristics of magnetic materials are very sensitive to temperature changes. Temperature increases can cause significant changes in parameters such as the coercivity and permeability of the core material, thereby affecting the loss level. Based on the Steinmetz equation, the present invention proposes a power function temperature correction equation, which takes into account the effect of temperature on core loss by introducing temperature-related terms. By introducing the temperature term, the effect of temperature on core loss can be taken into account, and the core loss can be predicted more accurately under different temperature conditions. Compared with the traditional Steinmetz equation, it can significantly improve the prediction accuracy, especially in high-frequency and high-power density application scenarios, where the temperature rise of the core is more significant and its advantages are more prominent.
[0023] The modified Steinmetz equation is as follows:
[0024]
[0025] Where T is the current internal temperature of the magnetic component in °C, and c is the correction factor. Parameter c controls the magnitude of the loss change with temperature.
[0026] In order to test the effect of the modified Steinmetz equation, it is compared with four other types of temperature correction equations (including linear temperature correction equation, quadratic temperature correction equation, logarithmic temperature correction equation and correction equation based on material properties). These correction equations also introduce temperature-related terms and consider the influence of temperature on core loss.
[0027] 1. Linear temperature correction equation
[0028] By adding a correction term linearly related to temperature T to the classic Steinmetz equation:
[0029]
[0030] This modified model assumes that the losses increase linearly with increasing temperature.
[0031] 2. Quadratic temperature correction equation
[0032] Assuming that the effect of temperature on core loss is nonlinear, consider the quadratic correction:
[0033]
[0034] This model shows that the effect of temperature on loss is not just linear, but also has a temperature square term. c1 and c2 are temperature correction coefficients used to adapt to more complex temperature changes.
[0035] 3. Logarithmic temperature correction equation
[0036] Assuming that the effect of temperature on core loss increases slowly, a logarithmic correction model can be used:
[0037]
[0038] This model assumes that the effect of temperature on loss increases gradually rather than rapidly and linearly, and is suitable for materials with mild temperature effects.
[0039] 4. Correction equations based on material properties
[0040] By considering the effect of temperature on material properties (such as resistivity or permeability), the modified equation can more comprehensively reflect the impact of temperature on core loss, thereby improving the accuracy of the prediction. This correction method is particularly suitable for temperature-sensitive magnetic materials because the resistivity and permeability of these materials change significantly with temperature, and the Steinmetz equation is indirectly modified through these physical quantities:
[0041]
[0042] This model considers the effect of temperature on loss through changes in material resistivity and is applicable to some temperature-sensitive magnetic materials.
[0043] Based on the same experimental data and fitting method, the corresponding parameters are obtained, as shown in Table 1.
[0044] Table 1
[0045]
[0046]
[0047] RMSE (root mean square error) and MSE (mean square error) are used as evaluation indicators, and the performance comparison of different equations is shown in Table 2.
[0048] Table 2
[0049]
[0050] From Table 2 and Figure 2 As shown, the modified Steinmetz equation proposed in the present invention has significantly improved the prediction accuracy compared with the traditional Steinmetz equation, especially in the case of larger loss values. The modified model can better capture the influence of external factors such as temperature on the core loss, thereby providing more accurate prediction results.
[0051] When predicting magnetic loss, the current temperature T is collected and the prediction result is output after calculation based on the modified Steinmetz equation.
[0052] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, a method for outputting a prediction result after performing a modified Steinmetz equation operation is implemented.
[0053] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.
Claims
1. A method for predicting magnetic loss of magnetic components based on the modified Steinmetz equation, characterized in that: The modified Steinmetz equation is: Where T is the internal temperature of the magnetic component, c is the correction factor, P is the core loss, f is the excitation frequency, and B m is the peak value of magnetic flux density; k1, ɑ1, and β1 are all material characteristic fitting parameters; the parameters k1, ɑ1, and β1 are given by the magnetic component manufacturer, or the correction coefficient c is obtained by fitting experimental data; When predicting magnetic loss, the temperature T is collected, and the prediction result is output after calculation based on the modified Steinmetz equation.
2. The method for predicting magnetic loss of a magnetic component based on the modified Steinmetz equation according to claim 1, characterized in that: When fitting parameters through experimental data, the experimental data is first preprocessed, including: cleaning the data, removing duplicate values and outliers, and ensuring the integrity and consistency of the data; normalizing the numerical data to eliminate the influence of different dimensions on the model; at the same time, encoding the categorical variables for application to the neural network model; then using the neural network fitting method and through nonlinear regression, the parameters required for fitting in the modified Steinmetz equation are obtained.
3. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for predicting magnetic loss of a magnetic component based on the modified Steinmetz equation according to claim 1 is implemented.