A Machine Learning-Based Intelligent Analysis Method for Steady-State Characteristics of DC-DC Converters

Abstract

This paper presents an intelligent analysis method for the steady-state characteristics of DC-DC converters based on machine learning, belonging to the field of DC-DC converters. It utilizes circuit simulation software to automatically obtain multiple sets of numerical solutions for the steady-state characteristics of the DC-DC converter under different driving strategies and system parameters. Then, it uses machine learning methods to fit the functional relationship between the steady-state characteristics of the DC-DC converter and the driving strategy and system parameters, thereby achieving intelligent analysis of the comprehensive steady-state characteristics of the DC-DC converter. The proposed method effectively overcomes the problem of difficulty in performing steady-state characteristic analysis of DC-DC converters considering parasitic parameters or containing multi-order resonant units using analytical methods. It also addresses the limitation of repeatedly performing simulations, which consumes a significant amount of time, when using simulation software to obtain the comprehensive steady-state characteristics of the DC-DC converter.

Description

Technical Field

[0001] This invention relates to the field of DC-DC converters, and in particular to a machine learning-based intelligent analysis method for steady-state characteristics of DC-DC converters, which can quickly and accurately analyze the comprehensive steady-state characteristics of complex DC-DC converters. Background Technology

[0002] When designing DC-DC converter products, it is necessary to first complete the steady-state characteristic analysis of the converter as the theoretical basis for the design. Currently, the common steady-state characteristic analysis methods for DC-DC converters mainly include analytical methods and numerical simulation methods. Among them, analytical methods are only applicable to solving differential equations within the second order. Therefore, it is difficult to complete the steady-state characteristic analysis for DC-DC converters that take into account the parasitic parameters of components and contain multi-order resonant units. To address this problem, researchers have proposed an analysis method for ultra-high frequency converters that integrates Kalman filtering technology ([1] Chen Yanfeng, Jiang Xinyi, Zhang Bo, Qiu Dongyuan, Xiao Wenxun, Xie Fan. Analysis method for ultra-high frequency converters that integrates Kalman filtering technology. CN 112507643 A) to quickly obtain the steady-state periodic analytical solution of the ultra-high frequency converter, thereby further solving the steady-state characteristics of the DC-DC converter. However, the implementation of this method is relatively complex, requiring the manual application of Kalman filtering technology, equivalent small parameter method, harmonic balance method, etc. In contrast, obtaining the steady-state characteristics of the DC-DC converter under given operating conditions and parameters through simulation software is relatively simple. However, each simulation can only obtain numerical solutions of component voltage and current under certain given operating conditions and parameters, making it difficult to know the comprehensive steady-state characteristics of the DC-DC converter. Therefore, how to achieve a simple and rapid comprehensive and accurate analysis of the steady-state characteristics of complex DC-DC converters still needs further research. Summary of the Invention

[0003] The purpose of this invention is to address the problem that existing technologies struggle to analyze the steady-state characteristics of DC-DC converters that take into account parasitic parameters or contain multi-order resonant units using analytical methods. This invention provides an intelligent analysis method for the steady-state characteristics of DC-DC converters based on machine learning. Through automated data acquisition and intelligent data analysis, it comprehensively understands and predicts the steady-state characteristics of DC-DC converters, achieving intelligent analysis of the full range of steady-state characteristics. Simultaneously, it overcomes the limitation of repeatedly performing simulations, which consumes a significant amount of time, when using simulation software to obtain the full range of steady-state characteristics of DC-DC converters.

[0004] To achieve the above-mentioned objectives, the present invention provides the following technical solution:

[0005] A machine learning-based intelligent analysis method for steady-state characteristics of DC-DC converters includes the following steps:

[0006] 1) Data acquisition: Using circuit simulation software, multiple sets of steady-state characteristic numerical solutions of the DC-DC converter under different driving strategies and system parameters are automatically obtained;

[0007] 2) Data preprocessing: Cleaning and organizing the data obtained from the simulation to ensure its accuracy and consistency;

[0008] 3) Machine learning model construction: The functional relationship between the steady-state characteristics of the DC-DC converter and the driving strategy and system parameters is fitted using machine learning methods;

[0009] 4) Model training and optimization: Use the preprocessed data from step 2) to train the machine learning model constructed in step 3), and improve the model's fitting accuracy and generalization ability by adjusting the model parameters and optimizing the algorithm;

[0010] 5) Model Validation and Application: The trained model is validated using test set data to evaluate its prediction accuracy and reliability; the validated model is applied to the steady-state characteristic analysis of actual DC-DC converters, and intelligent analysis of the comprehensive steady-state characteristics of DC-DC converters is achieved by inputting different driving strategies and system parameters.

[0011] In step 1), the circuit simulation software can be common power electronic circuit simulation software such as Matlab / Simulink, PSIM, PLECS, etc. The circuit simulation software is used to automatically modify parameters and realize batch simulation, thereby quickly generating a large number of numerical solutions about the steady-state characteristics of DC converter.

[0012] In step 1), the different driving strategies can be PWM, PFM, etc., and the turn-on and turn-off times of different switching transistors can be freely varied.

[0013] In step 1), the system parameters include switching frequency, switching duty cycle, input voltage, parameters of different components, and system parasitic parameters, etc.; through automated simulation, a large number of numerical solutions for the steady-state characteristics of the DC-DC converter are generated, including the average output voltage, output voltage ripple value, average inductor current, inductor current ripple value, RMS inductor current, average capacitor voltage, capacitor voltage ripple value, RMS switching current, average diode current, etc.

[0014] In step 2), the data preprocessing includes data cleaning, data organization, and feature extraction. Data cleaning is used to remove outliers and noise from the simulation data to ensure the accuracy and consistency of the data. Data organization is used to organize the simulation data into a format suitable for machine learning algorithms, such as CSV and Excel. Feature extraction is used to extract key parameters and indicators that have an important impact on the steady-state characteristics of the DC converter, such as the average value, effective value, and ripple of voltage and current, as input features for the machine learning model.

[0015] In step 3), the machine learning method is used to intelligently fit the functional mapping relationship between multiple input and output variables. Machine learning methods such as neural networks can be used to design a reasonable model structure, including the number of neurons in the input layer, hidden layer and output layer, activation functions, etc.

[0016] In step 3), the steady-state characteristics of the DC converter include electrical parameters such as the average value, effective value, and ripple of the voltage and current of each component.

[0017] Compared with the prior art, the present invention has the following outstanding technical effects and advantages:

[0018] This invention utilizes circuit simulation software to automatically obtain multiple sets of steady-state characteristic numerical solutions for DC-DC converters under different driving strategies and system parameters. It then uses machine learning methods to fit the functional relationship between the steady-state characteristics of the DC-DC converter and the driving strategy and system parameters, thereby achieving intelligent analysis of the comprehensive steady-state characteristics of the DC-DC converter. The automated simulation of the circuit simulation software and the rapid prediction of the machine learning algorithm significantly improve the efficiency of DC-DC converter steady-state characteristic analysis. The machine learning algorithm can intelligently fit complex functional relationships, thus achieving accurate prediction of the steady-state characteristics of the DC-DC converter. By covering a wide range of operating conditions and parameter combinations, it enables the analysis of the comprehensive steady-state characteristics of the DC-DC converter, providing strong support for optimized design and fault diagnosis. This invention effectively overcomes the problem of difficulty in using analytical methods to analyze the steady-state characteristics of DC-DC converters that consider parasitic parameters or contain multi-order resonant units, while also solving the limitation of repeatedly simulating and consuming a large amount of time when using simulation software to obtain the comprehensive steady-state characteristics of the DC-DC converter. Attached Figure Description

[0019] Figure 1 This invention relates to an intelligent analysis method for steady-state characteristics of DC-DC converters based on machine learning.

[0020] Figure 2 An example of a Buck converter that takes parasitic parameters into account;

[0021] Figure 3 for Figure 2 Training results for the average output voltage of the Buck converter;

[0022] Figure 4 for Figure 2 Training results for the average inductor current of the Buck converter;

[0023] Figure 5 for Figure 2 Average error rate of different electrical parameters in steady-state characteristic analysis of Buck converter. Detailed Implementation

[0024] To explain more clearly Figure 1 The present invention describes a machine learning-based intelligent analysis method for steady-state characteristics of DC-DC converters. The principle of this method will be explained in detail below with reference to the accompanying drawings and specific implementation methods. It is worth noting that the specific examples described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] This invention utilizes circuit simulation software to automatically obtain multiple sets of steady-state characteristic numerical solutions for DC-DC converters under different driving strategies and system parameters. By using machine learning methods, the functional relationship between the steady-state characteristics of the DC-DC converter and the driving strategy and system parameters is fitted, thereby realizing intelligent analysis of the comprehensive steady-state characteristics of the DC-DC converter.

[0026] The circuit simulation software, such as Matlab / Simulink, can automatically modify parameters to achieve batch simulations. The system parameters include switching frequency, switch duty cycle, input voltage, parameters of different components, and system parasitic parameters.

[0027] The steady-state characteristics of the DC-DC converter include electrical parameters such as the average value, effective value, and ripple of the voltage and current of each component.

[0028] The machine learning method described above can intelligently fit the functional mapping relationship between multiple input and output variables, such as neural networks and other machine learning methods.

[0029] like Figure 1 This paper presents an intelligent analysis method for steady-state characteristics of DC-DC converters based on machine learning. First, multiple sets of parameters for the DC-DC converter system are set, including switching frequency, duty cycle of the switching transistors, input voltage, parameters of different components, and parasitic parameters of the system. Then, circuit simulation software (such as the "sim" function in Matlab / Simulink) is used to perform batch simulations on the above-set parameters to obtain the average value, RMS value, ripple, and other electrical quantities of the corresponding components' voltage and current. These quantities are then used as the input and output of the machine learning algorithm for training.

[0030] by Figure 1 Taking the given neural network machine learning algorithm as an example, a neural network structure is constructed, consisting of an input layer, two hidden layers, and an output layer. The input layer is responsible for receiving external inputs, such as various parameters of the DC-DC converter, including switching frequency, duty cycle of the switching transistors, input voltage, parameters of different components, and system parasitic parameters, denoted as x1, x2, …, x… n The hidden layer is characterized by feature abstraction and nonlinear mapping, and is used to capture input data x1, x2, …, x n The complex relationships f1 and f2, etc., are contained within the layer; the output layer is used to output the final prediction results P1, P2, …, P.m Examples of parameters include the average value of the converter's output voltage or inductor current, and the ripple value. During the forward propagation phase, the input data is processed by multiple neurons in each layer, performing forward calculations layer by layer from the input layer to the output layer, and generating prediction results. During the backpropagation phase, based on the prediction results P1, P2, ..., P... m With expected results P'1, P'2, …, P' m The error between them is calculated using a gradient descent strategy, with the weights w in the network adjusted in the negative gradient direction of the target. ij The biases b1 and b2 are adjusted to minimize the mean squared error. The iterative forward and backward propagation processes are repeated until the network's mean squared error mse drops below a set threshold ε, or the maximum number of iterations is reached. At this point, the network's predicted output {P1, P2, …, P} is... m The result will be very close to the expected result {P'1, P'2, …, P'}. m After training, the neural network can intelligently predict the steady-state characteristics of a DC-DC converter. By inputting new DC-DC converter parameters, the neural network can quickly output prediction results. These prediction results can be used for performance analysis, optimization, and fault prediction of the DC-DC converter.

[0031] The following is a specific application example, applying the proposed machine learning-based intelligent analysis method for steady-state characteristics of DC-DC converters to a Buck converter that takes parasitic parameters into account. Figure 2 )middle.

[0032] 1. Parameter settings and simulation

[0033] Set the input voltage V of the converter in The voltage range is 36V to 66V, the inductance L is 15μH to 25μH, and the parasitic resistance R of the inductor is... L The resistance is 0.05Ω to 0.055Ω, the output capacitor C is 370nF to 470nF, and the parasitic resistance R of the capacitor is... C The resistance is 0.05Ω to 0.055Ω, the load resistance R is 6Ω to 16Ω, and the on-resistance R of the switching transistor S1 is... S The diode's forward voltage V is between 0.05Ω and 0.055Ω. F The voltage range is 0.4V to 1.2V, the duty cycle of the switching transistor varies from 0.3 to 0.8, and the switching frequency is 1MHz. Figure 2 In this context, V0 represents the output voltage; i L It represents electric current.

[0034] Within the above parameter range, 1000 sets of data were randomly selected, and batch simulations were performed on these 1000 sets of data using Matlab / Simulink simulation software to obtain the voltage and current values ​​of each component, thus obtaining 1000 sets of voltage and current values ​​for each component.

[0035] 2. Machine Learning Algorithm Training

[0036] We used 900 sets of data as the training set and the remaining 100 sets of data as the validation set.

[0037] Figure 3 exhibit Figure 2 The training results for the average output voltage of the Buck converter show that the goodness of fit of the training set, validation set, test set, and overall model is approximately 1, indicating that the neural network can fit and predict the changing trend of the output voltage very well. Figure 4 exhibit Figure 2 The training results for the average inductor current of the Buck converter show that the goodness of fit of the training set, validation set, test set, and the overall result is approximately 1, which also indicates that the neural network has high accuracy in predicting inductor current.

[0038] 3. Verification and Error Analysis

[0039] By using the remaining 100 sets of validation data for prediction, the average error rate of different electrical indices in the steady-state characteristic analysis of the Buck converter can be obtained. The results are then compared with the simulation results to calculate the average error rate of different electrical indices.

[0040] Figure 5 exhibit Figure 2 Average error rate of different electrical parameters (including RMS diode current I) in steady-state characteristic analysis of Buck converter Frms Average diode current I Favg , Effective value of switching transistor current I Srms Average value of switching transistor current I Savg Average inductor current I Lavg Inductor current RMS value I Lrms Inductor current peak-to-peak value I Lpk-pk Average output voltage U oavg Output voltage RMS value U orms Output voltage peak-to-peak value U opk-pk As can be seen, besides the peak-to-peak value of the inductor current I... Lpk-pk In addition, the calculation results for other voltage and current characteristics all had error rates within 0.1%. This indicates that the neural network has high accuracy and reliability in predicting the steady-state characteristics of the Buck converter.

[0041] This method can accurately predict key electrical parameters of a converter, such as output voltage and inductor current, providing strong support for converter design, optimization, and fault prediction. This method can be further extended to other types of DC-DC converters and more complex circuit systems, contributing to the intelligent development of power electronics.

[0042] In summary, the machine learning-based intelligent analysis method for steady-state characteristics of DC-DC converters proposed in this invention has the advantages of simple implementation and high accuracy. It can effectively overcome the shortcomings of traditional analytical or simulation methods, which is of great significance for the theoretical design of DC-DC converters and provides new ideas and methods for the intelligent analysis of power electronic systems.

[0043] The above embodiments are merely preferred embodiments of the present invention and should not be considered as limiting the scope of the present invention. All equivalent variations and improvements made within the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. A machine learning-based intelligent analysis method for steady-state characteristics of DC-DC converters, characterized in that... Includes the following steps: 1) Data Acquisition: Multiple sets of steady-state characteristic numerical solutions for the DC-DC converter under different driving strategies and system parameters are automatically acquired using circuit simulation software. The circuit simulation software used includes Matlab / Simulink, PSIM, and PLECS power electronic circuit simulation software. The circuit simulation software is used to automatically modify parameters and realize batch simulation, thereby quickly generating a large number of numerical solutions for the steady-state characteristics of the DC-DC converter. The different driving strategies use PWM or PFM, and the turn-on and turn-off times of different switches can be freely varied. The system parameters include switching frequency, switch duty cycle, input voltage, parameters of different components, and system parasitic parameters. The multiple sets of steady-state characteristic numerical solutions include the average value of output voltage, output voltage ripple value, average value of inductor current, inductor current ripple value, RMS value of inductor current, average value of capacitor voltage, capacitor voltage ripple value, RMS value of switch current, and average value of diode current. 2) Data preprocessing: Cleaning and organizing the data obtained from the simulation to ensure its accuracy and consistency; 3) Machine learning model construction: The functional relationship between the steady-state characteristics of the DC-DC converter and the driving strategy and system parameters is fitted using machine learning methods; 4) Model training and optimization: Use the preprocessed data from step 2) to train the machine learning model constructed in step 3), and improve the model's fitting accuracy and generalization ability by adjusting the model parameters and optimizing the algorithm; 5) Model Validation and Application: The trained model is validated using test set data to evaluate its prediction accuracy and reliability; the validated model is applied to the steady-state characteristic analysis of actual DC-DC converters, and intelligent analysis of the comprehensive steady-state characteristics of DC-DC converters is achieved by inputting different driving strategies and system parameters.

2. The intelligent analysis method for steady-state characteristics of DC-DC converters based on machine learning as described in claim 1, characterized in that... In step 2), the data preprocessing includes data cleaning, data organization, and feature extraction; data cleaning is used to remove outliers and noise from the simulation data to ensure the accuracy and consistency of the data. Data processing is used to organize simulation data into formats suitable for machine learning algorithms, including CSV and Excel. Feature extraction is used to extract key parameters and indicators that have an important impact on the steady-state characteristics of DC-DC converters, including the average value, RMS value, and ripple of voltage and current, as input features for machine learning models.

3. The intelligent analysis method for steady-state characteristics of DC-DC converters based on machine learning as described in claim 1, characterized in that... In step 3), the machine learning method is used to intelligently fit the functional mapping relationship between multiple input and output variables. The neural network machine learning method is used to design a reasonable model structure, including the number of neurons and activation functions of the input layer, hidden layer and output layer.

4. The intelligent analysis method for steady-state characteristics of DC-DC converters based on machine learning as described in claim 1, characterized in that... In step 3), the steady-state characteristics of the DC-DC converter include the average value, effective value, and ripple of the voltage and current of each component.

Images