IGBT Lifetime Extension via Adaptive Switching Frequency Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
IGBTs are sensitive to temperature changes, leading to reduced lifetime due to ineffective reactive temperature control, particularly in applications with varying external forces or power conversions, such as in compressors, cranes, and wind turbines.
Innovation Solution
A computer-implemented method using machine learning models to predict temperature changes in IGBTs based on current operating states and parameters, allowing for proactive adjustment of switching frequency and cooling system parameters to maintain the IGBT within a safe operating area, thereby reducing temperature fluctuations and extending its service life.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If reactive temperature control is used, then the control system is simple, but the IGBT lifetime is reduced due to ineffective temperature management
Solution Approach 1:
The system performs preliminary action by predicting future IGBT temperature changes using machine learning models before the temperature excursions occur. The predictive model analyzes current operating conditions and forecasts temperature trends, enabling proactive adjustment of switching frequency and cooling parameters to prevent excessive temperature variations that would reduce IGBT lifetime.
Solution Approach 2:
The system implements feedback by continuously monitoring actual IGBT temperature and comparing it with predicted temperature values. This feedback loop allows the control system to validate predictions, adjust the predictive model accuracy, and dynamically modify operating parameters to maintain optimal temperature ranges, thereby extending IGBT lifetime while managing system complexity.
2Productivity
If switching frequency is increased to improve productivity, then the operational output increases, but temperature changes intensify reducing IGBT lifetime
Solution Approach 1:
The system applies dynamics by making the switching frequency adjustable and adaptive rather than fixed. The predictive control dynamically modifies switching frequency based on forecasted temperature trends and actual temperature feedback, allowing the system to maintain high productivity when temperature conditions permit while preventing thermal damage when temperature excursions are predicted, thus extending IGBT lifetime.
Solution Approach 2:
The system changes parameters by dynamically adjusting switching frequency and cooling liquid flow rate based on predictive temperature models. When the model predicts excessive temperature rise, the system reduces switching frequency or increases cooling parameters to maintain safe operating temperatures, thereby protecting IGBT lifetime while minimizing impact on productivity.
3Temperature
If cooling liquid supply is increased to reduce temperature, then IGBT temperature decreases, but system complexity and energy consumption increase
Solution Approach 1:
The system applies partial action by providing cooling liquid supply only when and to the extent needed based on predictive temperature analysis. Rather than continuously maximizing cooling, the system adjusts cooling parameters proportionally to predicted temperature risks, reducing unnecessary energy consumption while maintaining adequate temperature control to extend IGBT lifetime.
4Measurement precision
If machine learning prediction is implemented, then temperature control accuracy improves, but device complexity increases
Solution Approach 1:
The system applies self-service by implementing self-learning capabilities where the machine learning model continuously improves its temperature prediction accuracy by learning from actual temperature measurements and operating condition patterns. This automated learning reduces the need for manual calibration and complex intervention, making the enhanced temperature control system progressively more efficient while maintaining manageable complexity.
Data Source
Figure 1
Figure 2
AI summary
A computer-implemented method (100) for operating an insulated-gate bipolar transistor, IGBT, comprising the steps of: • determining (110), by at least one machine learning model, based at least in part on the current operating state (1), and/or on current operating parameters (2), of the IGBT, a prediction of a change of the IGBT temperature (3) over a given temporal prediction horizon; • determining (120), based on this prediction (3) on the one hand and the current operating state (1), and/or on the current operating parameters (2), on the other hand, a modification (4) of the operating parameters of the IGBT, and/or of a cooling system that acts on the IGBT, that is likely to reduce the absolute value of the predicted change of the IGBT temperature; and • applying (130) this modification (4) to the operation of the IGBT, and/or for the cooling system.