A packaging method for high-current vehicle-mounted high-frequency inductor

Abstract

The present invention discloses a method for packaging a high-current vehicle-mounted high-frequency inductor, which relates to the technical field of power electronic device packaging, including: collecting existing inductor design parameters, material property data, and existing case data based on packaging requirements to construct a feature matrix; using a deep learning framework to establish an inductor packaging parameter optimization model that can process the feature matrix and predict the impact of packaging parameters on inductor performance; based on the difference between the preliminary simulation results and the prediction of the inductor packaging parameter optimization model, adjusting the model and optimizing the packaging design; making an inductor prototype according to the optimized packaging design, conducting actual tests, and formulating detailed packaging process specifications based on the test results. The present invention collects inductor data by constructing a feature matrix, the deep learning model accurately predicts the performance impact, the integrated simulation platform accelerates the optimization of packaging parameters, the closed-loop mechanism combines the algorithm to ensure efficient design, and the standardized testing and specifications improve product performance and competitiveness, creating a path for technological innovation in vehicle-mounted high-frequency inductors.

Description

Technical Field

[0001] The invention relates to the technical field of power electronic device packaging, in particular to a method for packaging a high-current vehicle-mounted high-frequency inductor. Background Art

[0002] With the rapid development of automotive electronic technology, especially the rise of new energy vehicles and autonomous driving technology, the complexity and power requirements of automotive electronic systems have increased dramatically. As a key component for energy storage and filtering, high-current automotive high-frequency inductors play a vital role in ensuring the stable operation of power electronic converters and improving system efficiency. Traditional inductor packaging methods mainly rely on engineers' experience and limited simulation methods, which are incapable of meeting the challenges of high-current and high-frequency applications.

[0003] Although the application of deep learning and integrated simulation technology in the field of inductor design has made certain progress, there are still several key problems. The existing technology has rough data preprocessing, which affects the prediction accuracy. It ignores physical laws when designing models, and the prediction deviation is large under high frequency and large current, which affects reliability. The integration efficiency of models and simulation platforms is low, which hinders automated design. The model adjustment strategy is extensive and lacks refined optimization, which affects the packaging performance. These problems directly restrict the accuracy and efficiency of inductor packaging design. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a high-current vehicle-mounted high-frequency inductor packaging method to solve the problems of rough data processing, model prediction deviation, inefficient simulation integration and extensive optimization strategy in the prior art.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a high-current automotive high-frequency inductor packaging method, which includes collecting existing inductor design parameters, material property data and existing case data based on packaging requirements, constructing a feature matrix, using a deep learning framework, establishing an inductor packaging parameter optimization model that can process the feature matrix and predict the impact of packaging parameters on inductor performance, developing an integrated simulation platform, interfacing the trained inductor packaging parameter optimization model with the simulation platform, using the inductor packaging parameter optimization model to generate packaging parameter recommendations, performing performance evaluation through the simulation platform to determine whether established standards are met, adjusting the model and optimizing the packaging design based on the difference between the preliminary simulation results and the predictions of the inductor packaging parameter optimization model, making an inductor prototype according to the optimized packaging design, conducting actual tests, and formulating detailed packaging process specifications based on the test results.

[0008] As a preferred solution of the high-current vehicle-mounted high-frequency inductor packaging method of the present invention, the specific steps of collecting existing inductor design parameters, material property data and existing case data based on packaging requirements are as follows:

[0009] Investigate the current needs of vehicle electronic systems, clearly define performance indicators, and collect design parameters and material property data through internal and external resources;

[0010] Collect and analyze the success and failure cases in high current and high frequency applications in the industry, compare the advantages and disadvantages of different design ideas, material selections and process flows, and extract key learning points and improvement directions;

[0011] Classify and integrate the survey and collected data to understand the current range and development trend in different application scenarios.

[0012] As a preferred solution of the high-current vehicle-mounted high-frequency inductor packaging method of the present invention, the specific steps of constructing the characteristic matrix are as follows:

[0013] Import the collected inductor design parameters, material property data and case data into a unified data management system for a preliminary integrity check;

[0014] First, outlier processing is performed, using box plot analysis to identify extreme values ​​in the data set and remove these outliers. Then, duplicate records in the data set are removed through careful comparison and inspection. Finally, consistency checks are performed to unify the units and formats of the data.

[0015] Based on the data after consistency check, the key features that directly reflect the inductor performance are extracted, and the contribution of each feature to the inductor performance and the vehicle is evaluated using the feature selection algorithm, and irrelevant features are eliminated;

[0016] The features after format unification and selection are organized into a table form, where each row represents an inductor design case and each column represents a feature vector, forming a feature matrix.

[0017] As a preferred solution of the high-current vehicle-mounted high-frequency inductor packaging method of the present invention, the deep learning framework is used to establish an inductor packaging parameter optimization model that can process the feature matrix and predict the impact of packaging parameters on inductor performance. The specific steps are as follows:

[0018] Construct an inductor package parameter optimization model, define a set of package parameters (P1, P2, ..., P), and use the inductor package parameter optimization model to predict the impact of the package parameters (P1, P2, ..., P) on the inductor performance Y;

[0019] Through the nonlinear relationship between the inductor performance Y and the packaging parameters, as well as the constraints of thermodynamics and electromagnetic principles, a physically inspired loss function L is designed, taking into account the prediction error and the consistency of the model prediction with the physical laws;

[0020] Use stochastic gradient descent (SGD) to minimize the comprehensive loss function L, iteratively update the model parameters θ, regularly monitor the performance on the validation set to tune the hyperparameters, evaluate the generalization ability of the model based on the test set, pay attention to the balance between prediction accuracy and physical consistency, and adjust the model architecture and hyperparameters through the Yeshiv optimization method to find the optimal configuration.

[0021] The integrated simulation platform is developed to interface the trained inductor packaging parameter optimization model with the simulation platform. The specific steps are as follows:

[0022] The design interface converts the output of the inductor package parameter optimization model (f) Seamlessly passed to the simulation platform, the interface includes, encapsulating the parameter vector [P1,P2,...,P m ] receiving end and predicted performance values The sending end;

[0023] Introducing parameter mapping function M(p i ), convert the package parameter pi into the input format required by the simulation platform;

[0024] The simulation software receives the package parameter vector [P1,P2,...,P m ], after M mapping, perform detailed electromagnetic field, thermal simulation and structural analysis, and output a set of simulation results Predicted value Through an adjustment function make corrections;

[0025] In the integrated simulation platform, the weight w2 of the physical consistency loss term PL is dynamically adjusted according to the difference between the model prediction and the simulation results;

[0026] An adaptive weight function w2(Δ) is designed to reflect the gap between the model prediction and the simulation results. As Δ decreases, the value of w2 gradually decreases, and as Δ increases, the value of w2 gradually increases.

[0027] The inductor package parameter optimization model is used to generate package parameter recommendations, and performance evaluation is performed through a simulation platform to determine whether the established standards are met. The specific steps are as follows:

[0028] Based on the project's goals, expected performance, and industry standards, a performance standard threshold T is defined, and then a comprehensive performance evaluation function G is defined to evaluate the consistency between the model prediction and the simulation results, while considering the fit between the performance goals and the physical laws;

[0029] The lower the G value, the higher the consistency between the model prediction and the simulation result, and at the same time, the deviation degree between the model prediction and the performance target is also smaller. Compare the G value with the preset performance standard threshold T. If G < T, the model prediction meets the established standard. If G ≥ T, the expression is that the established standard is not reached. According to the G value feedback, adjust the model parameters and optimize the packaging parameter vector [P1, P2,..., P m .

[0030] Based on the difference between the preliminary simulation result and the prediction of the inductor packaging parameter optimization model, adjust the model and optimize the packaging design. The specific steps are as follows:

[0031] Based on the difference between the preliminary simulation result and the model prediction, construct a difference sensitivity evaluation function D using a non - linear measurement method to quantify the difference between the two;

[0032] After obtaining the optimized packaging parameter vector, perform performance evaluation through the simulation platform again. Use the updated parameters for simulation and recalculate. If the value of the difference sensitivity evaluation function D decreases significantly and the G value is lower than the preset performance standard threshold T, it indicates that the packaging design meets the standard. If it does not meet the standard, return to the particle optimization algorithm for a new round of optimization iteration until the performance standard is reached.

[0033] Manufacture an inductor prototype according to the optimized packaging design, conduct actual tests, and formulate a detailed packaging process specification based on the test results. The specific steps are as follows:

[0034] Define a comprehensive optimization function H, an energy consumption efficiency function E, and a production cost function C, and combine them with the difference sensitivity evaluation function D to evaluate the global performance of the packaging design;

[0035] Set the upper limit I of the number of iterations and the convergence threshold T H , in each round of iteration, calculate the value of H(X k ), update the value of X according to the particle swarm optimization algorithm k until X k reaches the convergence threshold T H ;

[0036] Based on the optimized parameter vector X k conduct actual tests, manufacture an inductor prototype, and collect actual performance data. According to the test results, formulate a detailed packaging process specification.

[0037] In a second aspect, an embodiment of the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the high - current vehicle - mounted high - frequency inductor packaging method as described in the first aspect of the present invention is implemented.

[0038] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the high-current vehicle-mounted high-frequency inductor packaging method as described in the first aspect of the present invention is implemented.

[0039] The beneficial effects of the present invention are as follows: by constructing a feature matrix to systematically collect and organize inductor design parameters, material property data and case data, a solid data foundation is provided for the training of deep learning models, thereby realizing accurate prediction of factors affecting inductor performance. Subsequently, the establishment of a deep learning model and the development of an integrated simulation platform not only greatly improve the efficiency and accuracy of packaging parameter optimization, but also ensure the close fit of the design scheme at the theoretical and practical levels, significantly reducing the time cycle from design to mass production. The closed-loop performance evaluation and optimization mechanism, combined with sensitivity evaluation and particle swarm optimization algorithm, achieves a high degree of consistency between inductor performance and design goals. At the same time, through actual testing and the formulation of process specifications, the standardization and high efficiency of the production process are ensured, ultimately greatly improving product performance, stability and market competitiveness, and playing a positive role in promoting the development of the field of automotive electronic technology. This series of innovative steps together constitute an efficient and intelligent packaging design system, opening up a new path for the innovation and application of automotive high-frequency inductor technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0041] Figure 1 This is a flow chart of the high-current vehicle-mounted high-frequency inductor packaging method in Example 1.

[0042] Figure 2 This is a flow chart for formulating packaging process specifications in Example 1. DETAILED DESCRIPTION

[0043] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0044] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0045] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0046] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, provides a high-current vehicle-mounted high-frequency inductor packaging method, comprising the following steps:

[0047] S1. Based on packaging requirements, collect existing inductor design parameters, material property data, and existing case data to build a feature matrix.

[0048] Going a step further, investigate the current needs of in-vehicle electronic systems, clearly define performance indicators (such as operating voltage, rated current, Q value, etc.), and collect design parameters and material property data through internal materials and external resources (including scientific research papers and technical reports);

[0049] Collect and analyze the success and failure cases in high current and high frequency applications in the industry, compare the advantages and disadvantages of different design ideas, material selections and process flows, and extract key learning points and improvement directions;

[0050] Classify and integrate the survey and collected data to understand the current range and development trend in different application scenarios.

[0051] Import the collected inductor design parameters, material property data and case data into a unified data management system for a preliminary integrity check;

[0052] First, outlier processing is performed, using box plot analysis to identify extreme values ​​in the data set and remove these outliers. Then, through careful comparison and inspection, duplicate records in the data set are eliminated to ensure the uniqueness and accuracy of the information. Finally, consistency checks are implemented to unify the units and formats of the data.

[0053] Based on the data after consistency check, key features that directly reflect the inductor performance are extracted, including inductor size, number of coil turns, material thermal conductivity, etc. The contribution of each feature to the inductor performance and the vehicle is evaluated using the feature selection algorithm, and irrelevant features are eliminated;

[0054] The features after format unification and selection are organized into a table form, where each row represents an inductor design case and each column represents a feature vector, forming a feature matrix.

[0055] It should also be noted that the box plot analysis method for identifying extreme values ​​in a data set mainly includes three detailed steps: first, by drawing a box plot to intuitively display the data distribution, the location of the outliers is determined based on the five-number summary (minimum value, Q1, median, Q3, maximum value); second, the causes of the outliers are analyzed and decisions are made on how to deal with them, which includes deleting, correcting, or retaining the outliers; finally, after implementing the treatment measures, the cleaning effect is ensured by revalidating the data, thereby improving data quality and the accuracy of subsequent analysis. This process helps to systematically manage and reduce anomalies and noise in the data set, laying a solid foundation for deeper data exploration and model building.

[0056] S2. Using a deep learning framework, an inductor packaging parameter optimization model is established that can process feature matrices and predict the impact of packaging parameters on inductor performance.

[0057] Furthermore, an inductor package parameter optimization model is constructed, and a set of package parameters (P1, P2, ..., P) is defined. The inductor package parameter optimization model is used to predict the influence of the package parameters (P1, P2, ..., P) on the inductor performance Y. The expression is:

[0058] Y pr =f(P1,P2,...,P;θ);

[0059] Among them, θ represents the parameter set of the model, Y pr is the predicted inductance performance value, P m represents the mth encapsulation parameter, and f represents a mapping function;

[0060] Through the nonlinear relationship between the inductor performance Y and the packaging parameters, as well as the constraints of thermodynamics and electromagnetic principles, a physically inspired loss function L is designed, taking into account the prediction error and the consistency of the model prediction with the physical laws. The expression is:

[0061]

[0062] Among them, MSE represents mean square error, PL is a physical consistency loss term, is the predicted value adjusted by physical constraints, Y tr Represents the actual inductance performance index value, Y pr is the inductor performance index value predicted by the model, w1 is used to control the influence of the MSE term, and w2 is used for the proportion of the PL term;

[0063] Use stochastic gradient descent (SGD) to minimize the comprehensive loss function L, iteratively update the model parameters θ, regularly monitor the performance on the validation set to tune the hyperparameters, evaluate the generalization ability of the model based on the test set, pay attention to the balance between prediction accuracy and physical consistency, and adjust the model architecture (number of layers, number of nodes) and hyperparameters (learning rate, regularization term) through the Ye Si optimization method to find the optimal configuration.

[0064] It should also be noted that SGD adopts the form of mini-batch, that is, each update is not based on a single sample, nor on the entire data set, but on a small batch of randomly selected samples. This can reduce the computational overhead and improve the stability of the update. It is iterated multiple times until the preset number of iterations is reached or the stopping condition is met (such as the gradient change is less than a certain threshold). The expression is:

[0065]

[0066] Among them, θ t are the model parameters at time step t, η is the learning rate, is the loss function L with respect to the parameter θt in the sample (x i ;y i ), the gradient on x i is the input feature vector of a specific sample, y i is the same as the input feature vector x i The associated correct answer or “true” label;

[0067] S3. Develop an integrated simulation platform to interface the trained inductor packaging parameter optimization model with the simulation platform.

[0068] Furthermore, the design interface converts the output of the inductor package parameter optimization model (f) Seamlessly passed to the simulation platform, the interface includes, encapsulating the parameter vector [P1,P2,...,P m ] receiving end and predicted performance values The sending end;

[0069] Introducing parameter mapping function M(p i ), converting the package parameter pi into the input format required by the simulation platform to match the input requirements of the simulation software;

[0070] The simulation software receives the package parameter vector [P1,P2,...,P m ], after M mapping, perform detailed electromagnetic field, thermal simulation and structural analysis, and output a set of simulation results Predicted value Through an adjustment function make corrections;

[0071] In the integrated simulation platform, the weight w2 of the physical consistency loss term PL is dynamically adjusted according to the difference between the model prediction and the simulation result;

[0072] Design an adaptive weight function w2(Δ), where Δ = ∣A(Y pr ) - Y si ∣, which reflects the gap between the model prediction and the simulation result. As Δ decreases, the value of w2 also gradually decreases, encouraging the model to learn predictions closer to the physical laws. As Δ increases, the value of w2 also gradually increases, emphasizing the importance of physical consistency and prompting the model to pay more attention to following the physical laws.

[0073] It should also be noted that when developing the integrated simulation platform, a sophisticated interface was designed to ensure the smooth exchange of information between the inductor package parameter optimization model and the simulation platform. The parameter mapping function M(p i ) plays a bridging role, converting the package parameter vector [P1, P2,..., P m into a format recognized by the simulation platform, ensuring an accurate connection from the model to the simulation environment. Subsequently, the simulation platform uses these parameters for in-depth electromagnetic field, thermal, and structural analyses to generate simulation results.

[0074] S4. Use the inductor package parameter optimization model to generate package parameter suggestions, and conduct performance evaluations through the simulation platform to determine whether the established standards are met.

[0075] Furthermore, based on the project's goals, expected performance, and industry standards, define a performance standard threshold T, and then define a comprehensive performance evaluation function G, which is used to evaluate the consistency between the model prediction and the simulation result, while considering the degree of fit between the performance target and the physical laws. The expression is:

[0076]

[0077] Among them, is the corrected predicted value vector, is the simulation result vector, is the performance target vector, w1 is the performance index weight vector, w2(Δ) is the physical consistency loss weight vector, N is the number of performance indicators, and i is the index variable.

[0078] The lower the value of G, the higher the consistency between the model prediction and the simulation result, and at the same time, the smaller the deviation between the model prediction and the performance target. Compare the value of G with the preset performance standard threshold T. If G < T, the model prediction meets the established standards. If G ≥ T, the expression indicates that the established standards have not been reached. According to the feedback of the G value, adjust the model parameters and optimize the package parameter vector [P1, P2,..., P m ;

[0079] It should also be noted that the range of G values ​​is theoretically from 0 to infinity, but in practical applications, due to the existence of the performance target vector, the G value will be limited to a limited area. A G value close to 0 indicates that the model prediction is highly consistent with the simulation results, and is also close to the performance target, indicating that the model performance is excellent. As the G value increases, the inconsistency between the model prediction and the simulation results increases, or the degree of deviation from the performance target increases, indicating the need to further optimize the model or adjust the parameters.

[0080] S5. Based on the difference between the preliminary simulation results and the predictions of the inductor package parameter optimization model, adjust the model and optimize the package design.

[0081] Furthermore, based on the difference between the preliminary simulation results and the model prediction, a difference sensitivity evaluation function D is constructed using a nonlinear measurement method to quantify the difference between the two. The expression is:

[0082]

[0083] Among them, Y i,pr represents the corrected model prediction value vector, Y si represents the simulation result vector, N is the number of performance indicators, and Y" pr is the predicted value vector, Y i si represents the i-th element in the simulation value sequence;

[0084] In this function Partially ensures sensitivity to small differences, while It ensures exponential penalty for large differences, thus achieving comprehensive coverage of difference sensitivity assessment;

[0085] Based on the feedback of the difference sensitivity evaluation function D, the particle swarm optimization algorithm is introduced to adjust the model parameters and optimize the package parameter vector [P1, P2, ..., P m ];

[0086] Set the particle position vector X in PSO K =[P1,P2,...,P m ], velocity vector V k And the fitness function F, expressed as:

[0087] F(x k )=D(Y″ pr ,Y si )+α·G(Y″ pr ,Y si ,w1,w2(Δ));

[0088] Where, is the balance factor, F(X k) is the entire evaluation function;

[0089] After obtaining the optimized package parameter vector, the performance is evaluated again through the simulation platform, and the updated parameters are used for simulation to recalculate D(Y″ pr ,Y si )and If the value of the difference sensitivity evaluation function D is significantly reduced and the G value is lower than the preset performance standard threshold T, it means that the packaging design meets the standard. If it does not meet the standard, it returns to the particle optimization algorithm for a new round of optimization iteration until the performance standard is met.

[0090] S6. Make an inductor prototype according to the optimized packaging design, conduct actual tests, and formulate detailed packaging process specifications based on the test results.

[0091] Furthermore, the comprehensive optimization function H, energy efficiency function E and production cost function C are defined and combined with the heterosensitivity evaluation function D to evaluate the global performance of the package design. The expression is:

[0092] H(x k )=D(Y″ pr ,Y si )+α·E(x k ,P e )+β·C(x k ,P c );

[0093] Among them, E(X k ,P e ) is the energy efficiency function per unit output in the packaging process, P e is the energy efficiency parameter matrix, C(X k ,P c ) is the energy efficiency function per unit output in the packaging process, and β is used to adjust the importance of the production cost function relative to other components;

[0094] Set the upper limit of iteration number I and convergence threshold T H , in each iteration, calculate H(X k ) value, and update X according to the particle swarm optimization algorithm k The value of X k Reaching the convergence threshold T H ;

[0095] Based on the optimized parameter vector X k Conduct actual tests, manufacture inductor prototypes, and collect actual performance data. Based on the test results, develop detailed packaging process specifications.

[0096] It should also be noted that the particle position update expression is:

[0097]

[0098] The particle velocity update expression is:

[0099]

[0100] Among them, X k (t) represents the position vector of particle k at the tth iteration, X k (t+1) It means that at the t+1th iteration, V k (t) represents the velocity vector of particle k and the position vector of particle k at the t+1th iteration, w is the inertia weight, c1 and c2 are learning factors, r1 and r2 are random numbers between [0,1], pt k is the historical optimal position of particle k, and gt is the historical optimal position of the entire particle group;

[0101] This embodiment also provides a computer device, which is suitable for the case of a high-current vehicle-mounted high-frequency inductor packaging method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the high-current vehicle-mounted high-frequency inductor packaging method proposed in the above embodiment.

[0102] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0103] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for packaging a high-current vehicle-mounted high-frequency inductor proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0104] In summary, the present invention: The present invention integrates key data by constructing a feature matrix, lays the foundation for the model, ensures data quality and analysis depth, establishes an inductor packaging parameter optimization model, combines the physical loss function with the stochastic gradient descent algorithm, realizes accurate prediction and physical compliance, develops an integrated simulation platform, dynamically adjusts physical consistency, enhances prediction reliability, performance evaluation and standard judgment mechanism, finely adjusts model parameters and packaging design, combines actual testing with comprehensive optimization functions to ensure high performance and economy, and the entire system significantly improves the efficiency, accuracy and reliability of inductor design, provides technical support for the development of high-performance components of high-frequency inductors in automotive electronic systems, and promotes the performance and stability of automotive electronic systems.

[0105] Example 2, referring to Table 1, is the second example of the present invention. To further verify the advancement of the present invention, experimental simulation data of a high-current vehicle-mounted high-frequency inductor packaging method are provided.

[0106] Three different performance indicators were selected, and each indicator was set with different operating voltage, rated current, Q value, and temperature rise.

[0107] First, the inductor design parameters, material data and cases are collected to construct a feature matrix.

[0108] Secondly, a deep learning framework is used to model and predict the impact of packaging parameters, taking physical constraints into account during training.

[0109] Next, a simulation platform was developed to interface with the model, evaluate the inductor performance, adjust the model parameters, and optimize the design.

[0110] Finally, the inductor prototype is made, actual testing is carried out, and packaging process specifications are formulated to ensure stable and efficient performance.

[0111] The details are shown in Table 1:

[0112] Table 1 Experimental record table

[0113]

[0114] By comparing the test data, it can be seen that under the same working voltage, the rated current of the technology of the present invention is significantly higher than that of the existing technologies A and B, which is increased by 50% and 20% respectively. In addition, the Q value has also been improved, from 45.0 and 42.0 to 50.0 respectively, which means that the quality factor of the inductor is better and the energy loss is lower. In terms of temperature rise, the temperature rise of the technology of the present invention is only 28.0°C, which is lower than 32.5°C and 35.0°C of the existing technologies A and B, which shows that under high power operation, the inductor heat dissipation performance of this technology is better. Finally, although the rated current and Q value have been improved, the packaging volume of the technology of the present invention has been reduced from 150cm 3 and 180cm 3 Down to 140cm 3 , reflecting higher space utilization and lightweight design.

[0115] The high-current vehicle-mounted high-frequency packaging method of the present invention has shown superior characteristics to the existing technology in terms of rated current, Q value, temperature rise and packaging volume, fully demonstrating its innovation and advantages. This is due to the precise prediction capability of the deep learning model and the high-performance evaluation of the simulation platform, which enables the optimization of packaging parameters, thereby significantly improving the performance and reliability of the inductor in high-frequency and high-current environments.

[0116] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for packaging a high-current vehicle-mounted high-frequency inductor, characterized in that: including Based on the packaging requirements, collect the existing inductance design parameters, material property data, and existing case data to construct a feature matrix Adopt a deep learning framework to establish an inductance packaging parameter optimization model that can process the feature matrix and predict the impact of packaging parameters on inductance performance Develop an integrated simulation platform and interface the trained inductance packaging parameter optimization model with the simulation platform Use the inductance packaging parameter optimization model to generate packaging parameter suggestions, conduct performance evaluation through the simulation platform, and determine whether the established standards are met Based on the differences between the preliminary simulation results and the predictions of the inductance packaging parameter optimization model, adjust the model and optimize the packaging design Fabricate an inductance prototype according to the optimized packaging design, conduct actual tests, and formulate detailed packaging process specifications based on the test results The specific steps for constructing the feature matrix are as follows Import the collected inductance design parameters, material property data, and case data into a unified data management system for preliminary integrity checking Perform outlier processing. Use box plot analysis to identify extreme values in the dataset and remove the outliers. Then, through careful comparison and checking, eliminate duplicate records in the dataset. Finally, perform consistency verification to unify the units and formats of the data Based on the data after consistency verification, extract the key features that directly reflect the inductance performance, and use feature selection algorithms to evaluate the contribution of each feature to the inductance performance and the vehicle Organize the features after format unification and selection into a table form, with each row representing an inductance design case and each column representing a feature vector to form a feature matrix The specific steps for using the inductance packaging parameter optimization model to generate packaging parameter suggestions, conduct performance evaluation through the simulation platform, and determine whether the established standards are met are as follows Based on the project objectives, expected performance, and industry standards, define a performance standard threshold T, and then define a comprehensive performance evaluation function G to evaluate the consistency between the model prediction and the simulation results, while considering the compliance of the performance target with the physical laws The lower the value of G, the higher the consistency between the model prediction and the simulation results, and the smaller the deviation of the model prediction from the performance target. Compare the value of G with the preset performance standard threshold T. If G < T, the model prediction meets the established standards. If G ≥ T, it means that the established standards are not reached. According to the feedback of the G value, adjust the model parameters and optimize the packaging parameter vector [P1,P2,...,P m ]; The specific steps for adjusting the model and optimizing the packaging design based on the differences between the preliminary simulation results and the predictions of the inductance packaging parameter optimization model are as follows Based on the differences between the preliminary simulation results and the model predictions, construct a difference sensitivity evaluation function D using a non-linear measurement method to quantify the differences between the two After obtaining the optimized packaging parameter vector, conduct performance evaluation through the simulation platform again. Use the updated parameters for simulation and recalculate. If the value of the difference sensitivity evaluation function D decreases significantly and the value of G is lower than the preset performance standard threshold T, it indicates that the packaging design meets the standards. If not, return to the particle optimization algorithm for a new round of optimization iteration until the performance standards are met 2. The high-current vehicle-mounted high-frequency inductor packaging method according to claim 1, characterized in that: The specific steps of collecting existing inductor design parameters, material property data and existing case data based on packaging requirements are as follows: Investigate the current needs of vehicle electronic systems, clearly define performance indicators, and collect design parameters and material property data through internal and external resources; Collect and analyze the success and failure cases in high current and high frequency applications in the industry, compare the advantages and disadvantages of different design ideas, material selections and process flows, and extract key learning points and improvement directions; Classify and integrate the survey and collected data to understand the current range and development trend in different application scenarios.

3. The high-current vehicle-mounted high-frequency inductor packaging method according to claim 2, characterized in that: The deep learning framework is used to establish an inductor packaging parameter optimization model that can process the feature matrix and predict the impact of packaging parameters on inductor performance. The specific steps are as follows: Construct an inductor package parameter optimization model, define a set of package parameters (P1, P2, ..., P), and use the inductor package parameter optimization model to predict the impact of the package parameters (P1, P2, ..., P) on the inductor performance Y; Through the nonlinear relationship between the inductor performance Y and the packaging parameters, as well as the constraints of thermodynamics and electromagnetic principles, a physically inspired loss function L is designed, taking into account the prediction error and the consistency of the model prediction with the physical laws; Use stochastic gradient descent (SGD) to minimize the comprehensive loss function L, iteratively update the model parameters θ, regularly monitor the performance on the validation set to tune the hyperparameters, evaluate the generalization ability of the model based on the test set, pay attention to the balance between prediction accuracy and physical consistency, and adjust the model architecture and hyperparameters through the Yeshiv optimization method to find the optimal configuration.

4. The high-current vehicle-mounted high-frequency inductor packaging method according to claim 3, characterized in that: The integrated simulation platform is developed to interface the trained inductor packaging parameter optimization model with the simulation platform. The specific steps are as follows: The design interface converts the output of the inductor package parameter optimization model (f) Seamlessly passed to the simulation platform, the interface includes, encapsulating the parameter vector [P1,P2,...,P m ] receiving end and predicted performance values The sending end; Introducing parameter mapping function M(p i ), convert the package parameter pi into the input format required by the simulation platform; The simulation software receives the package parameter vector [P1,P2,...,P m ], after M mapping, perform detailed electromagnetic field, thermal simulation and structural analysis, and output a set of simulation results Predicted value Through an adjustment function A make corrections; In the integrated simulation platform, the weight w2 of the physical consistency loss term PL is dynamically adjusted according to the difference between the model prediction and the simulation results; An adaptive weight function w2(Δ) is designed to reflect the gap between the model prediction and the simulation results. As Δ decreases, the value of w2 gradually decreases, and as Δ increases, the value of w2 gradually increases.

5. The high-current vehicle-mounted high-frequency inductor packaging method according to claim 4, characterized in that: The inductor prototype is made according to the optimized packaging design, actual testing is carried out, and detailed packaging process specifications are formulated based on the test results. The specific steps are as follows: Define the comprehensive optimization function H, energy efficiency function E and production cost function C, and combine them with the heterosensitivity evaluation function D to evaluate the global performance of the package design; Set the upper limit of iteration number I and convergence threshold T H , in each iteration, calculate H(X k ) value, and update X according to the particle swarm optimization algorithm k The value of X k Reaching the convergence threshold T H ; Based on the optimized parameter vector X k Conduct actual tests, manufacture inductor prototypes, and collect actual performance data. Based on the test results, develop detailed packaging process specifications.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the high-current vehicle-mounted high-frequency inductor packaging method described in any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the high-current vehicle-mounted high-frequency inductor packaging method described in any one of claims 1 to 5 are implemented.

Images