Porsche Engineering has developed an artificial intelligence-controlled soft switching system that reduces power transistor switching losses in electric vehicle inverters by 70 to 95 percent. The technology, currently in advanced simulation testing, processes real-time vehicle data to optimize transistor switching timing, potentially increasing EV range by a high single-digit percentage while shrinking inverter volume by 20 to 50 percent.
Highlights
- AI-controlled Zero Voltage Switching (ZVS) reduces inverter switching losses by 70-95%, extending EV range
- Pre-trained algorithms process load, torque, and temperature data to calculate optimal switching times in real time
- Inverter volume reduced 20-50% through lower cooling requirements and elimination of filter components
- Software-based implementation allows integration into existing control units as a plug-in solution
Understanding Inverter Losses
Electric vehicle inverters experience two primary types of energy losses: line losses and switching losses. Line losses result from residual resistance inherent to power transistors and cannot be influenced by circuit design. Switching losses occur during transitions between on and off states, when voltage remains present while current flows through the transistor.
Higher switching frequencies improve alternating current quality but increase switching losses proportionally. This creates a fundamental efficiency trade-off that conventional hard switching approaches cannot resolve.
Zero Voltage Switching Approach
Porsche Engineering selected Zero Voltage Switching over Zero Current Switching for several technical reasons. ZVS produces lower losses in silicon-carbide and gallium-nitride transistors commonly used in EVs. The approach delivers better efficiency at higher frequencies with reduced electromagnetic interference.
“Around the power transistors, we place an additional electronic circuit consisting of transistors, coils, and capacitors,” said Souhaib Touati, Specialist Project Engineer Function & Software Development at Porsche Engineering. “This inverter topology has long been referred to as the auxiliary resonant commutated pole, or ARCP.”
AI Enables Real-Time Optimization
The innovation centers on using artificial intelligence to control the ARCP circuit. A pre-trained algorithm processes several dozen live measurements from the vehicle to calculate optimal switching times in fractions of a second.
“An electric vehicle operates under constantly changing loads, which is why soft switching has not yet been applied in any other way due to the wide variety of rapidly changing operating conditions,” said Volker Reber, Senior Manager Function & Software Development at Porsche Engineering. “Our algorithm predicts the optimum moments for controlling the ARCP, even under changing conditions.”
The engineering team is investigating two AI methodologies:
- Recursive neural networks: Characterized by high predictive accuracy and high performance
- Reinforcement learning: Increases calculation speed, suited for demanding real-time requirements
Technical and Commercial Benefits
The efficiency gains extend beyond range improvements. Lower switching losses reduce heat generation in the pulse inverter, decreasing cooling requirements and enabling more compact components. Various filter components found in existing circuits can be eliminated, optimizing printed circuit board layouts.
Hard switching places significant stress on power transistors, reducing service life. AI-based soft switching alleviates this wear mechanism, potentially improving inverter durability.
Software-Based Implementation
Porsche Engineering plans to offer AI-based soft switching as a complete software solution once development concludes. The system integrates into existing control units through software libraries.
“We will thereby offer OEMs and Tier-1 suppliers a big advantage: AI-based soft switching is implemented purely via software libraries and can therefore be integrated into existing control units in the same way as a plug-in,” said Touati. “The cost of modifying the hardware should be relatively low based on existing designs.”
The technology would be best implemented during model updates or new vehicle development programs.
Source: Porsche Newsroom
Sign up for our popular daily email to catch all the latest EV news!
