Power Module RUL Prediction Using Adaptive Switch Voltage Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining the remaining useful life of power modules, such as those containing field effect transistors (FETs) and insulated gate bipolar transistors (IGBTs), are inadequate in accurately predicting the health and lifespan of these components, particularly due to temperature variations and lack of real-time adaptation.
Innovation Solution
A regression model is applied by a processor to analyze voltage measurements from power switches, adapting over time to characterize the specific power module, using inputs like drain-to-source voltage (VDS) to predict the state of health (SOH) and remaining useful life (RUL), with temperature compensation and wire break detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional fixed regression models are used for predicting power module lifespan, then the model structure is simple, but the prediction accuracy deteriorates due to temperature variations and lack of real-time adaptation
Solution Approach 1:
The patent transforms the static regression model into a dynamic adaptive model that continuously updates its parameters based on real-time operational data. The model adapts to changing temperature conditions and power module degradation patterns, allowing it to maintain high prediction accuracy throughout the module's lifespan rather than degrading over time.
Solution Approach 2:
The system implements a feedback mechanism where actual operational data from the power module (including temperature readings and electrical parameters) is continuously fed back into the regression model. This feedback loop enables the model to learn from real-world performance and adjust its parameters accordingly, improving prediction accuracy while accounting for environmental variations.
2Reliability
If real-time adaptive regression modeling is implemented to improve prediction accuracy, then the prediction reliability improves, but the computational complexity and processing requirements increase
Solution Approach 1:
The patent focuses on adapting specific regression parameters (coefficients and intercepts) rather than fundamentally changing the entire model structure. By adjusting only the necessary parameters based on temperature conditions and operational data, the system achieves improved reliability while minimizing the increase in computational complexity. The core regression framework remains intact, reducing processing requirements compared to complete model retraining.
Data Source
AI summary
A system may comprise a power module including a power switch, a driver circuit configured to control an ON state and an OFF state of the power switch, and a processor configured to control the driver circuit. The processor may be configured to receive a voltage associated with the power switch in the ON state and determine a state of health (SOH) for the power switch based on the voltage and a regression model associated with the power switch.