DAB Converter Parameter Estimation Using Physics-Informed Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating circuit parameters of DAB converters face challenges due to high computational burdens, data sparsity, and limited generalization capabilities, particularly in complex DC-AC-AC-DC systems, and existing neural network-based approaches require extensive training data and suffer from overfitting.
Innovation Solution
A physics-informed neural network (PINN) is employed to estimate DAB converter parameters by deriving time-domain differential equations, establishing a physical connection between power signals and parameters, and using a data-driven network to update weights and biases, incorporating both data and physical information for accurate estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional physical information-based parameter estimation methods are used, then measurement precision can be achieved, but device complexity and computational burden increase significantly
Solution Approach 1:
The patent replaces conventional physical information-based estimation methods with a data-driven neural network approach. The neural network learns parameter estimation directly from input-output data patterns, substituting complex physical modeling and mathematical calculations with a trained computational model that achieves comparable or superior accuracy with reduced implementation complexity.
Solution Approach 2:
The patent creates a virtual copy of the DAB converter system through a neural network model that replicates the input-output behavior. This digital twin approach allows parameter estimation without physically implementing complex measurement systems, as the neural network captures the system's characteristics through training data.
2Ease of operation
If data-driven parameter estimation methods using AI tools are employed, then ease of operation improves, but manufacturing precision deteriorates due to poor performance in data-sparse fields
Solution Approach 1:
The patent performs preliminary action by training the neural network offline using simulated data that covers the full operating range of the DAB converter. This pre-training phase creates a robust model that can handle data-sparse scenarios during actual operation, as the network has already learned from comprehensive training data including edge cases and varying operating conditions.
Solution Approach 2:
The patent changes the parameter representation by using normalized operating parameters (phase shift ratio, voltage ratios) as inputs to the neural network. This parameter transformation allows the model to generalize across different operating conditions and maintain accuracy even with limited measurement data during deployment.
3Productivity
If existing neural network-based parameter estimation methods are applied, then productivity increases through automated estimation, but reliability decreases due to overfitting risks and poor generalization capability
Solution Approach 1:
The patent implements feedback mechanisms during the training process by using loss functions that compare neural network predictions with ground truth values from simulations. This continuous feedback allows the model to learn from errors and generalize better to unseen operating conditions, reducing overfitting risks while maintaining fast inference speed during deployment.
Solution Approach 2:
The patent creates a universal neural network model that can estimate multiple parameters (inductance, capacitance, resistance, voltage ratios) simultaneously from the same input data. This multi-functional approach improves reliability by using all available information comprehensively rather than training separate models, while maintaining high productivity through a single unified estimation process.
4Measurement precision
If physical information-based methods with additional sensors are used, then measurement precision improves, but device complexity and cost increase
Solution Approach 1:
The patent enables the DAB converter system to estimate its own parameters using only its existing operational data (input voltage, output voltage, phase shift). The neural network processes normally available operational signals to derive parameter information, allowing the system to self-diagnose and self-monitor without requiring additional external sensors or measurement hardware.
Data Source
AI summary
A method for estimating circuit parameters of DAB converters based on a physics-informed neural network is disclosed, and belongs to the field of circuit parameter estimation technology. By inputting a small amount of inductor current and output voltage data from the DAB converter, the above method is used to achieve high-precision circuit parameter estimation with strong robustness and good generalization capability, with the impact of noise and different modulation strategies on the estimation accuracy taken into consideration.