A parameter extraction method of a magnetic core loss model considering frequency and temperature effects

Abstract

The application discloses a parameter extraction method of a magnetic core loss model considering frequency and temperature influence, comprising the following steps: 1) obtaining magnetic flux density waveforms and corresponding magnetic core loss data of a magnetic material under different frequency and temperature working conditions; 2) constructing a magnetic core loss prediction model based on an improved generalized Steinmetz equation, and extracting IGSE model parameters under different working conditions by using an individualized chaotic JAYA optimization algorithm; 3) constructing a nonlinear mapping function of the IGSE loss parameters with respect to frequency and temperature; and 4) searching and optimizing the mapping function parameters by using the individualized chaotic JAYA optimization algorithm. The method can accurately depict the nonlinear characteristics of the magnetic core loss parameters changing with frequency and temperature, and the individualized chaotic parameters and learning update mechanism can effectively weaken the convergence prematurity and search stagnation phenomenon in the algorithm, so that high-precision prediction of the magnetic core loss under different working conditions is realized.

Description

Technical Field

[0001] This invention relates to the field of magnetic core loss modeling and prediction technology, and in particular to a parameter extraction method for a magnetic core loss model that considers the effects of frequency and temperature. Background Technology

[0002] As power electronics technology continues to advance towards higher frequencies, higher power densities, and higher efficiency, more stringent requirements are being placed on the miniaturization, high performance, and high reliability of magnetic components in power electronic devices. Inductors, transformers, and other magnetic components are key parts of power electronic systems, and their size, efficiency, and thermal characteristics largely determine the overall system performance. Core loss is one of the main forms of energy loss generated by magnetic components under alternating magnetic fields, directly affecting the temperature rise, efficiency, and lifespan of magnetic components. Therefore, accurate modeling and prediction of core loss is a crucial foundation for the optimized design of magnetic components and the efficient operation of power electronic systems.

[0003] To address the modeling and calculation of magnetic core losses, scholars both domestically and internationally have proposed various analytical methods. Existing magnetic core loss calculation models can be broadly categorized into three types: loss separation models, physical models based on hysteresis mechanisms, and empirical formula models. Loss separation models typically decompose magnetic core losses into components such as hysteresis loss, eddy current loss, and additional losses, possessing certain physical meaning. However, these models often rely on numerous material parameters and complex experimental measurements, primarily applied to loss analysis in electromagnetic devices such as motors, and have limitations in engineering applications within the power electronics field. Physical models based on hysteresis mechanisms, such as the Jiles-Atherton model and the Preisach model, can accurately describe the magnetization process and hysteresis characteristics of magnetic materials. However, their complex model structures, numerous parameters, and computationally intensive parameter identification processes make them unsuitable for the rapid design and simulation requirements of power electronic systems.

[0004] In contrast, empirical formula models, due to their simplicity and high computational efficiency, have become the most widely used method for predicting magnetic core losses in engineering practice. Among them, the Steinmetz equation and its improved forms model core losses using macroscopic quantities such as frequency and magnetic flux density, making them well-suited to datasheets provided by magnetic material manufacturers and easy to use directly in the power electronics system design process. To improve the applicability of empirical formula models under non-sinusoidal excitation conditions, researchers have proposed improved Steinmetz equations, generalized Steinmetz equations, and improved generalized Steinmetz equations based on the traditional Steinmetz equation. The improved generalized Steinmetz equation, by incorporating time-varying information of the magnetic flux density waveform, has improved the prediction accuracy of core losses under non-sinusoidal excitation conditions to a certain extent, and has therefore received widespread attention and application in the field of power electronics.

[0005] However, existing core loss models based on Steinmetz-type empirical formulas typically assume that model parameters remain constant within a certain operating range. These parameters are often extracted from experimental data under sinusoidal excitation and a limited frequency range. When the operating frequency, temperature, or magnetic flux density waveform changes, fixed parameters cannot accurately reflect the nonlinear characteristics of core loss as a function of operating conditions, leading to a significant decrease in prediction accuracy over a wide frequency range or under multiple temperature conditions. Furthermore, in practical engineering applications, different frequency ranges often require the extraction of a separate set of model parameters. The lack of continuity between these parameters can easily lead to discontinuities at the boundaries of frequency or temperature ranges, thus affecting the stability and reliability of core loss prediction results.

[0006] In extracting parameters for magnetic core loss models, traditional methods often employ deterministic optimization techniques such as least-squares fitting. These methods are highly sensitive to initial conditions and data quality, and are prone to getting trapped in local optima in multi-parameter, strongly nonlinear problems. With the development of intelligent optimization techniques, some studies have attempted to introduce swarm intelligence optimization algorithms to search for parameters in magnetic core loss models, aiming to improve the accuracy and robustness of parameter extraction. However, most existing parameter extraction methods still primarily focus on fitting parameters under single or discrete operating conditions, making it difficult to comprehensively characterize the continuous evolution of model parameters with frequency and temperature variations. The efficiency of parameter extraction and the model's generalization ability still need further improvement.

[0007] In summary, existing core loss prediction models and parameter extraction methods still have certain shortcomings under complex operating conditions such as wide frequency and multiple temperatures, and it is still difficult to simultaneously achieve prediction accuracy, model continuity and engineering practicality. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides a parameter extraction method for a magnetic core loss model that considers the effects of frequency and temperature. This method can accurately characterize the relationship between model parameters and frequency and temperature while maintaining the clarity of the physical meaning of the magnetic core loss model, and achieve efficient and stable extraction of its parameters for magnetic core loss prediction, thereby meeting the needs of modern power electronic systems for high-performance modeling of magnetic components.

[0009] The objective of this invention is achieved as follows: a method for extracting parameters of a magnetic core loss model that considers the effects of frequency and temperature, comprising the following steps:

[0010] Step 1) Obtain the magnetic flux density waveform and corresponding core loss data of the magnetic material under different frequencies and temperatures;

[0011] Step 2) Based on the obtained magnetic flux density waveform and core loss data, a core loss model is constructed using the improved generalized Steinmetz equation, and the core loss model parameters under different frequency and temperature conditions are extracted using an optimization algorithm.

[0012] Step 3) Construct a nonlinear mapping function of the core loss model parameters with respect to frequency and temperature, and determine the parameters of the nonlinear mapping function to be extracted;

[0013] Step 4) Use an optimization algorithm to search for the parameters to be extracted in the nonlinear mapping function, and substitute the obtained mapping function parameters into the core loss model to establish a core loss prediction model that considers the effects of frequency and temperature.

[0014] Furthermore, the core loss model described in step 2) is specifically as follows:

[0015] (1)

[0016] in, Represents the loss coefficient parameter. , The exponential parameter represents the magnetic flux density, and the model parameters are used to reflect the loss characteristics of magnetic materials under different operating frequencies and magnetic flux density conditions.

[0017] Furthermore, the nonlinear mapping function described in step 3) is expressed as:

[0018]

[0019]

[0020]

[0021] in, ~ , ~ , ~ These are the parameters of the mapping function to be extracted.

[0022] The nonlinear mapping function is used to characterize the relationship between the loss coefficients k, α, and β in the improved generalized Steinmetz equation core loss model and the operating frequency f and temperature T.

[0023] Furthermore, the optimization algorithm described in steps 2) and 4) is the individualized chaotic JAYA optimization algorithm. The individualized chaotic JAYA optimization algorithm is based on the optimal solution guidance and worst solution suppression update mechanism of the JAYA optimization algorithm. During the algorithm iteration process, individualized control parameters are set for different search individuals in the optimization algorithm population to achieve differentiated search updates.

[0024] Meanwhile, a chaotic mechanism is introduced to generate or update random factors for algorithm initialization and iterative updates, thereby perturbing the position information of the search individual. The position information of the search individual is iteratively updated until the preset termination condition is met, thus obtaining the optimal solution of the core loss model parameters or nonlinear mapping function parameters.

[0025] Furthermore, in the individualized chaotic JAYA optimization algorithm, individualized control parameters are updated or generated through a chaotic mechanism to simulate the process of differentiated learning; the chaotic control parameters used to adjust the update behavior of the search individual are generated or updated using a logistic mapping, and the iterative formula is as follows:

[0026] (5)

[0027] in, This represents the chaos control parameters corresponding to the j-th search entity during the m-th iteration. These are the chaos control parameters updated by the chaos mapping, used to adjust the update weights of the search individuals.

[0028] Furthermore, in the individualized chaotic JAYA optimization algorithm, for non-optimal search individuals, before performing chaotic mapping on the control parameters, a Gaussian random perturbation is first introduced into the control parameters corresponding to the non-optimal search individuals to form perturbed intermediate control parameters. These intermediate control parameters serve as input variables for the chaotic mapping.

[0029] (6)

[0030] For non-optimal search entities, their position vectors are updated according to the following update formula based on the optimal solution-guided and worst solution-suppressed update mechanism:

[0031]

[0032] in, This represents the current position of the j-th search individual in the i-th dimension; and These represent the position information of the best and worst individuals in the population in the i-th dimension during the current iteration. , is a random factor in the interval (0,1); W is a weight adjustment factor used to enhance the suppression of the worst solution.

[0033] Furthermore, in the individualized chaotic JAYA optimization algorithm, for the optimal search individual obtained in the current iteration, its corresponding chaotic control parameters are independently updated through a logistic mapping, and the update relationship is as follows:

[0034] (8)

[0035] in, and These represent the chaotic control parameters corresponding to the optimal search individual in the m-th and m+1-th iterations, respectively;

[0036] For the optimal search individual obtained in the current iteration, without introducing Gaussian random perturbation, the position vector of the optimal search individual is updated based on the control parameters generated by the chaotic mapping. The update method is expressed as follows:

[0037] (9).

[0038] Furthermore, the weight adjustment factor W is used to adjust the dynamic weight parameter that controls the intensity of the search individual's move away from the worst solution update, and its specific update formula is as follows:

[0039]

[0040] in, , These represent the best and worst fitness values ​​under the objective function in each iteration.

[0041] Furthermore, in the process of optimizing the core loss model parameters and nonlinear mapping function parameters using the individualized chaotic JAYA optimization algorithm, the normalized root mean square error (RMS) between the predicted and experimental core loss values, and between the mapping reconstruction parameters and the baseline optimal parameters, are used as the optimization objective functions, and their expressions are as follows:

[0042] (11)

[0043] (12)

[0044] in, This is the predicted core loss value corresponding to the nth sample point obtained based on the core loss prediction model. The experimental measurement value of the core loss corresponding to the nth sample point; is the average value of the experimental measurements of core loss; N is the number of sample points used in the optimization. For the IGSE model, the parameters are k, α, or β; The mapping reconstruction parameter value corresponding to the i-th working condition; Let be the baseline optimal parameter value corresponding to the i-th working condition; t represents the average of the baseline optimal parameter values; t represents the number of sample operating conditions.

[0045] Furthermore, during the iterative process of the individualized chaotic JAYA optimization algorithm, a learning update mechanism is introduced for the chaotic control parameters corresponding to the search individual. This learning update mechanism, when the search individual fails to achieve a better fitness after updating, updates the parameters with probability p. r Reset individualization parameter C m To achieve a better fitness, the control parameters are adaptively adjusted by weighting and fusing the control parameters in the current iteration with the candidate control parameters. The update method satisfies the following relationship:

[0046] (13)

[0047] in, This represents the chaos control parameters corresponding to the j-th search individual in the m-th iteration; This represents the candidate control parameters generated based on the chaotic mapping; η is the learning factor, used to adjust the weight ratio between historical control parameters and candidate control parameters during the update process.

[0048] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes a core loss model that considers the influence of frequency and temperature. By constructing the model parameters in the improved generalized Steinmetz equation as a nonlinear mapping function with respect to the operating frequency and temperature, it can effectively characterize the nonlinear characteristics of the core loss parameters as a function of frequency and temperature. This avoids the problem of limited applicability caused by the use of fixed parameters in the traditional core loss model, and improves the prediction accuracy and adaptability of the core loss prediction model under wide frequency and multi-temperature conditions.

[0049] This invention proposes an individualized chaotic JAYA optimization algorithm by introducing individualized control parameters and a learning update mechanism. This enhances the global search capability and search diversity of the optimization algorithm in the parameter search process, effectively reduces the problem of traditional optimization algorithms easily getting trapped in local optima in complex parameter spaces, and improves the accuracy of the extraction results of core loss model parameters and nonlinear mapping function parameters. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0051] Figure 1 This is a schematic diagram of the process for extracting parameters of a magnetic core loss model based on an individualized chaotic JAYA optimization algorithm according to an embodiment of the present invention.

[0052] Figure 2 This is a scatter plot of the predicted core loss values ​​of a core loss model (based on the individualized chaotic JAYA optimization algorithm) considering frequency and temperature, according to an embodiment of the present invention.

[0053] Figure 3 This is a convergence curve of the IGSE parameter search process of different optimization algorithms in one embodiment of the present invention at 25℃ and 446kHz.

[0054] Figure 4 This is a scatter plot of the predicted core loss values ​​of a conventional IGSE core loss prediction model according to an embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] like Figure 1 The parameter extraction method for a magnetic core loss model considering the effects of frequency and temperature, as shown, includes the following steps:

[0057] Step 1): Obtain the magnetic flux density waveform and corresponding core loss experimental data of the magnetic material under different frequencies and temperatures. In this embodiment, data of N87 type material is obtained from the MagNet database. The number of sample points is the number of triangular wave samples in the N87 material data, i.e., the number of samples is set to 13190.

[0058] Step 2): Based on the acquired magnetic flux density waveform and core loss data, a core loss prediction model is constructed using the improved generalized Steinmetz equation (IGSE). Based on the objective function, an optimization algorithm is used to extract the core loss model parameters under different frequency and temperature conditions. The IGSE model is as follows:

[0059] (1)

[0060] The parameters of the IGSE model include k, α, and β, where k represents the loss coefficient parameter, and α and β represent exponential parameters related to magnetic flux density.

[0061] The objective function is the normalized root mean square error (RMS) between the predicted and experimentally measured core loss values, specifically:

[0062] (2)

[0063] in, This is the predicted core loss value corresponding to the nth sample point obtained based on the core loss prediction model. The experimental measurement value of the core loss corresponding to the nth sample point; is the average value of the experimental measurements of core loss; N is the total number of samples, in this embodiment N=13190.

[0064] Step 3): Construct a nonlinear mapping function for the core loss model parameters with respect to operating frequency and temperature, and determine the mapping parameters to be extracted from the nonlinear mapping function; in this embodiment, the function is constructed for the parameters in the IGSE model, and the specific function is as follows:

[0065]

[0066]

[0067]

[0068] in, The unit is kHz. ~ , ~ , ~ These are the parameters of the mapping function to be extracted.

[0069] Step 4): Construct the objective function, based on the normalized root mean square error (RMS):

[0070] (6)

[0071] Where θ is k, α, or β; The mapping reconstruction parameter value corresponding to the i-th working condition; Let be the baseline optimal parameter value corresponding to the i-th working condition; t represents the average of the baseline optimal parameter values; t represents the number of sample operating conditions, which is 76 in this embodiment.

[0072] Step 5): Optimize the parameters extracted from the core loss model that considers the effects of frequency and temperature based on the optimization algorithm. This invention employs a swarm intelligence optimization algorithm, which distributes a certain number of individuals within the search space and allows these individuals to evolve according to a specific mechanism, thereby gradually approaching the optimal value within the search space and achieving the goal of optimization.

[0073] The optimization algorithm is an individualized chaotic JAYA optimization algorithm. The individualized chaotic JAYA optimization algorithm is based on the optimal solution guidance and worst solution suppression update mechanism of the JAYA optimization algorithm. By setting individualized control parameters for different search individuals in the optimization algorithm population, a differentiated search update strategy is realized. At the same time, a chaotic mechanism is introduced to generate or update random factors in the algorithm to enhance the ergodicity and randomness of the search process and avoid the search process from getting trapped in local optima.

[0074] Step 5-1): Set the maximum number of iterations, the population size of the optimization algorithm, and randomly generate the positions of individuals in the population of the optimization algorithm;

[0075] Step 5-2): Calculate the fitness value of each individual in the population of the optimization algorithm based on the objective function;

[0076] Step 5-3): Update or generate individualized control parameters through a chaotic mechanism to simulate the process of differentiated learning. The iterative formula for updating or generating individualized chaotic control parameters is as follows:

[0077] (7)

[0078] in, This represents the chaos control parameters corresponding to the j-th search entity during the m-th iteration. These are the chaos control parameters updated by the chaos mapping, used to adjust the update weights of the search individuals.

[0079] For non-optimal search individuals, before performing chaotic mapping on the control parameters, a Gaussian random perturbation is first introduced into the control parameters corresponding to the non-optimal search individuals to form perturbed intermediate control parameters. These intermediate control parameters serve as input variables for the chaotic mapping.

[0080]

[0081] For the optimal search individual obtained in the current iteration, its corresponding chaotic control parameters are updated independently through a logistic mapping, and the update relationship is as follows:

[0082] (9)

[0083] in, and These represent the chaotic control parameters corresponding to the optimal search individual in the m-th and m+1-th iterations, respectively;

[0084] During the iterative process of the individualized chaotic JAYA optimization algorithm, a learning and updating mechanism is introduced for the chaotic control parameters corresponding to the search individual. This mechanism achieves adaptive adjustment of the control parameters by weighted fusion of the control parameters in the current iteration and the candidate control parameters. The update method satisfies the following relationship:

[0085] (10)

[0086] in, This represents the chaos control parameters corresponding to the j-th search individual in the m-th iteration; This represents the candidate control parameters generated based on the chaotic mapping; η is the learning factor, used to adjust the weight ratio between historical control parameters and candidate control parameters during the update process. For individuals that do not achieve a better fitness, there is a probability that the individual's control parameters will be reset.

[0087] Step 5-4): Use the optimal solution-guided and worst-solution-suppression update mechanism in conjunction with the individualized control parameters to update the position information of different individuals in the population of the optimization algorithm.

[0088] For non-optimal search entities, their position vectors are updated according to the following update formula based on the optimal solution-guided and worst solution-suppressed update mechanism:

[0089]

[0090] in, This represents the current position of the j-th search individual in the i-th dimension; and represents the position information of the best and worst individuals in the population in the i-th dimension during the current iteration; r1 and r2 are random factors in the interval (0,1); W is a weight adjustment factor used to enhance the suppression of the worst solution.

[0091] The weight adjustment factor W is a dynamic weight parameter used to adjust the update intensity of individual searches that are far from the worst solution. Its value is set according to the search state during the algorithm iteration process, so as to enhance the global search capability in the early stage of the algorithm iteration and improve the convergence accuracy in the later stage of the algorithm iteration. Its specific update formula is as follows:

[0092]

[0093] Where, f(X) b f(X) w ) represent the best and worst fitness of each iteration under the objective function.

[0094] For the optimal search individual obtained in the current iteration, without introducing Gaussian random perturbation, the position vector of the optimal search individual is updated based on the control parameters generated by the chaotic mapping. The update method is expressed as follows:

[0095]

[0096] Step 5-5): Calculate the current iteration count. If the current iteration count is greater than the maximum iteration count, proceed to step 5-7; otherwise, proceed to step 5-6.

[0097] Steps 5-6): Increment the iteration count by 1, then go to step 5-2).

[0098] Steps 5-7): The result of extracting the parameters of the core loss model considering the effects of frequency and temperature is the position vector of the optimal individual in the population of the optimization algorithm.

[0099] The above embodiments enable accurate identification of core loss model parameters that consider the effects of frequency and temperature. A scatter plot of the predicted core loss values ​​for the core loss model considering frequency and temperature (based on the individualized chaotic JAYA optimization algorithm) is shown below. Figure 2 As shown in Table 1, the predicted core loss value output by the model has a strong linear correlation with the measured value, indicating that the established model can accurately characterize the law of core loss change with operating conditions. The RMS error calculated by formula (3) is 5.72%.

[0100] Table 1. Results of core loss prediction based on different optimization algorithms (N87 material)

[0101]

[0102] Based on the Individualized Chaotic JAYA Optimization Algorithm, the average RMS values ​​obtained from core loss prediction after extracting model parameters using the Chaotic JAYA Optimization Algorithm, Kepler Optimization Algorithm, Moss Growth Optimization Algorithm, and Harris Eagle Optimization Algorithm were 6.10, 7.21, 10.25, 12.05, and 17.45, respectively, with standard deviations of 0.315, 0.656, 1.910, 1.953, and 4.022. The Individualized Chaotic JAYA Optimization Algorithm demonstrates superior convergence accuracy and stability compared to the other four algorithms. This invention introduces an individualized control operator into the Individualized Chaotic JAYA Optimization Algorithm, enabling different search individuals in the population to have differentiated search step sizes and update strategies. This enhances the dynamic adaptive capability of the population search process, effectively improves the diversity and synergy of population evolution, and significantly reduces the premature convergence phenomenon where the algorithm easily gets trapped in local optima in complex parameter spaces. Meanwhile, chaotic mappings possess stronger ergodicity and random perturbation capabilities compared to traditional uniform random distributions. This invention generates and updates individualized chaotic factors based on chaotic mappings to drive key random parameters in the JAYA algorithm, enabling individual search entities to achieve an adaptive balance between global exploration and local exploitation during iteration. Furthermore, this invention introduces a learning update mechanism to provide feedback updates to individual chaotic factors, allowing individuals to inherit and reinforce effective search behavior, thereby improving the algorithm's convergence speed and optimization accuracy. Experimental results verify the advancement of the method presented in this invention.

[0103] The convergence curves of different optimization algorithms in the IGSE parameter search process at 25℃ and 446kHz are shown below. Figure 3 As shown. (Through) Figure 3 It can be seen that ICJAYA not only has a faster convergence speed under this condition, but also achieves stronger local exploration and escape from local optima in the later stages of iteration, thereby obtaining the lowest fitness value, demonstrating better optimization accuracy and better convergence stability. This result verifies that the individualized chaos mechanism and learning update strategy have a significant effect on improving the exploration ability of the JAYA algorithm and avoiding premature convergence.

[0104] The scatter plot of the core loss prediction values ​​of the traditional IGSE core loss prediction model is shown below. Figure 4 As shown. (Through) Figure 4It can be seen that when using the traditional IGSE fixed-parameter model for core loss prediction, the scatter distribution of the predicted and measured values ​​exhibits significant dispersion, with a large number of sample points deviating from the ideal reference line. This indicates that the fixed-parameter model is difficult to accurately describe the variation of core loss under different operating conditions. The RMS value of the traditional IGSE core loss prediction model is 33.98%, which is higher than the prediction error of any algorithm in the embodiments of the core loss model considering frequency and temperature proposed in this invention. In contrast, it is evident that the core loss model proposed in this invention, whose parameters vary with frequency and temperature, constructs a nonlinear mapping relationship between the loss parameters and frequency and temperature, enabling the model parameters to adaptively adjust according to different operating conditions. This effectively improves the consistency between the predicted results and the actual loss, significantly reduces the error level, and demonstrates stronger adaptability, accuracy, and engineering application value.

[0105] The experimental results above show that, compared with the traditional IGSE fixed parameter model, the core loss model proposed in this invention, which considers the effects of frequency and temperature, can more accurately describe the variation law of core loss under different operating conditions.

[0106] The above description of the embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. It should be noted that those skilled in the art can make several improvements and modifications to the present invention without departing from the principles of the present invention, and these improvements and modifications also fall within the protection scope of the claims of the present invention.

Claims

1. A method for extracting parameters of a magnetic core loss model considering the effects of frequency and temperature, characterized in that, Includes the following steps: Step 1) Obtain the magnetic flux density waveform and corresponding core loss data of the magnetic material under different frequencies and temperatures; Step 2) Based on the obtained magnetic flux density waveform and core loss data, a core loss model is constructed using the improved generalized Steinmetz equation, and the core loss model parameters under different frequency and temperature conditions are extracted using an optimization algorithm. Step 3) Construct a nonlinear mapping function of the core loss model parameters with respect to frequency and temperature, and determine the parameters of the nonlinear mapping function to be extracted; Step 4) Use an optimization algorithm to search for the parameters to be extracted in the nonlinear mapping function, and substitute the obtained mapping function parameters into the core loss model to establish a core loss prediction model that considers the effects of frequency and temperature.

2. The parameter extraction method for a core loss model considering the effects of frequency and temperature according to claim 1, characterized in that, The core loss model described in step 2) is as follows: (1)； in, Represents the loss coefficient parameter. , The exponential parameter represents the magnetic flux density, and the model parameters are used to reflect the loss characteristics of magnetic materials under different operating frequencies and magnetic flux density conditions.

3. The parameter extraction method for a magnetic core loss model considering the effects of frequency and temperature according to claim 1, characterized in that, The nonlinear mapping function mentioned in step 3) is expressed as: ； ； ； in, ~ , ~ , ~ These are the parameters of the mapping function to be extracted; The nonlinear mapping function is used to characterize the relationship between the loss coefficients k, α, and β in the improved generalized Steinmetz equation core loss model and the operating frequency f and temperature T.

4. The parameter extraction method for a magnetic core loss model considering the effects of frequency and temperature according to claim 1, characterized in that, The optimization algorithm described in steps 2) and 4) is the Individualized Chaotic JAYA Optimization Algorithm. The Individualized Chaotic JAYA Optimization Algorithm is based on the optimal solution guidance and worst solution suppression update mechanism of the JAYA Optimization Algorithm. During the algorithm iteration process, individualized control parameters are set for different search individuals in the optimization algorithm population to achieve differentiated search updates. Meanwhile, a chaotic mechanism is introduced to generate or update random factors for algorithm initialization and iterative updates, thereby perturbing the position information of the search individual. The position information of the search individual is iteratively updated until the preset termination condition is met, thus obtaining the optimal solution of the core loss model parameters or nonlinear mapping function parameters.

5. The parameter extraction method for a magnetic core loss model considering the effects of frequency and temperature according to claim 4, characterized in that, In the individualized chaotic JAYA optimization algorithm, individualized control parameters are updated or generated through a chaotic mechanism to simulate the process of differentiated learning. The chaotic control parameters used to adjust the update behavior of the search individual are generated or updated using a logistic mapping, and the iterative formula is as follows: (5)； in, This represents the chaos control parameters corresponding to the j-th search entity during the m-th iteration. These are the chaos control parameters updated by the chaos mapping, used to adjust the update weights of the search individuals.

6. The parameter extraction method for a core loss model considering the effects of frequency and temperature according to claim 4, characterized in that, In the individualized chaotic JAYA optimization algorithm, for non-optimal search individuals, before performing chaotic mapping on the control parameters, a Gaussian random perturbation is first introduced into the control parameters corresponding to the non-optimal search individuals to form perturbed intermediate control parameters. These intermediate control parameters serve as input variables for the chaotic mapping. (6)； For non-optimal search entities, their position vectors are updated according to the following update formula based on the optimal solution-guided and worst solution-suppressed update mechanism: (7)； in, This represents the current position of the j-th search individual in the i-th dimension; and These represent the position information of the best and worst individuals in the population in the i-th dimension during the current iteration. , is a random factor in the interval (0,1); W is a weight adjustment factor used to enhance the suppression of the worst solution.

7. The parameter extraction method for a core loss model considering the effects of frequency and temperature according to claim 4, characterized in that, In the individualized chaotic JAYA optimization algorithm, for the optimal search individual obtained in the current iteration, its corresponding chaotic control parameters are independently updated through a logistic mapping, and the update relationship is as follows: (8)； in, and These represent the chaotic control parameters corresponding to the optimal search individual in the m-th and m+1-th iterations, respectively; For the optimal search individual obtained in the current iteration, without introducing Gaussian random perturbation, the position vector of the optimal search individual is updated based on the control parameters generated by the chaotic mapping. The update method is expressed as follows: (9)。 8. The parameter extraction method for a magnetic core loss model considering the effects of frequency and temperature according to claim 6, characterized in that, The weight adjustment factor W is a dynamic weight parameter used to adjust the update strength of the search individual away from the worst solution. Its specific update formula is as follows: (10)； in, , These represent the best and worst fitness values ​​under the objective function in each iteration.

9. The parameter extraction method for a core loss model considering the effects of frequency and temperature according to claim 4, characterized in that, In the process of optimizing the parameters of the core loss model and the nonlinear mapping function using the individualized chaotic JAYA optimization algorithm, the normalized root mean square error (RMS) between the predicted and experimental core loss values, and between the mapping reconstruction parameters and the baseline optimal parameters, are used as the optimization objective functions, and their expressions are as follows: (11)； (12)； in, This is the predicted core loss value corresponding to the nth sample point obtained based on the core loss prediction model. The experimental measurement value of the core loss corresponding to the nth sample point; is the average value of the experimental measurements of core loss; N is the number of sample points used in the optimization. For the IGSE model, the parameters are k, α, or β; The mapping reconstruction parameter value corresponding to the i-th working condition; Let be the baseline optimal parameter value corresponding to the i-th working condition; t represents the average of the baseline optimal parameter values; t represents the number of sample operating conditions.

10. The parameter extraction method for a core loss model considering the effects of frequency and temperature according to claim 4, characterized in that, In the iterative process of the individualized chaotic JAYA optimization algorithm, a learning update mechanism is introduced for the chaotic control parameters corresponding to the search individual. This learning update mechanism, when the search individual fails to achieve a better fitness after updating, updates with probability p... r Reset individualization parameter C m To achieve a better fitness, the control parameters are adaptively adjusted by weighting and fusing the control parameters in the current iteration with the candidate control parameters. The update method satisfies the following relationship: (13)； in, This represents the chaos control parameters corresponding to the j-th search individual in the m-th iteration; This represents the candidate control parameters generated based on the chaotic mapping; η is the learning factor, used to adjust the weight ratio between historical control parameters and candidate control parameters during the update process.

Images