Finding the right parts is an uphill task for power electronics engineers, whose choice of battlefield ranges from SiC wafers to soporific MOSFETs and diodes – and now, SBDs etc, AI-assisted tools should help them wade through this blizzard of components with an eye towards high performance specs for electric vehicles – not just cars, but sub-your-car-sized truck and their charging stations too – photovoltaic inverters, high-temperature industrial gear, metallurgy applications, and more.
Why AI?
“There typically are multiple competing objectives to take into account as power engineers devise systems for their desired applications,” said Robert Hsu, Director of Marketing at GeneSic, whose AI tools help optimize the selection of semiconductors. “These target applications might include electric vehicles, photovoltaic inverters, high-temperature industrial equipment, and, metallurgical operations. Given the diversity of power device requirements, human-based selection typically exhibits a relatively high tolerance for error, and this selection can become a slow process.AI systems can analyze these enormous data sets to determine:
Electrical specs (voltage, current, temp rating)
Material characteristics (breakdown voltage, thermal conductivity)
Manufacturer quality and price info
Historical failure data and reliability models
AI can sift through all this to pick devices that will help boost the effect on overall system performance goals but minimize the chance of susceptible device failures affecting the overall system carve-out.Silicon Carbide Components
Silicon carbide wafers and substrates are the heart of high-power semiconductors, and using appropriately tutored machine learnings on the devices for these components allows users to:
characterize wafer quality on defects (density and uniformity)
characterize wafers on carrier mobility and doping concentration
characterize wafers on thermal conductivity and mechanical stabilitySiC semiconductors: MOSFETs, diodes, SBDs
SiC MOSFETs are frequently employed in high-frequency, high-voltage application based on achieving lower switching losses than with their silicon brethren, and AI systems can help the designer home in on the most appropriate kind of such MOSFET based on:
On-resistance (RDS(on)) and speed of switching;
Maximum drain-source voltage;
Thermal performance in the application environment;SiC diodes and SBDs (Schottky Barrier Diodes) also benefit from AI selection to aid in reducing reverse recovery loss and improving efficiency in EV inverter system application, photovoltaic inverter, and industrial power converter.AI in selection of electrical protection devices
Another example of user substitution of clever AI systems is in determining the correct electrical-protection devices, fuses or circuit breaker or smart protection module, for a custom power system carve-out.The AI-assisted tool will weigh:
Load characteristics & fault current levels
Environmental and operational parameters
Coordination of upstream and downstream protection
Minimizing the cost of downtime, minimizing damage from lat trip or fuse blow, and protecting human life in a high-voltage industrial system.
Even commercial delivery systems of lower rated voltage but critical service for power delivery are protected with AI assistance.
Practical Benefits to Power Engineers
Using these examples of actual implementing of AI-assisted component selection from suppliers, we hope to provide insight into the following few major practical benefits of AI approaches to power engineers:
Time savings on design – the power engineer in complex systems where they must get power system protection selections right is allowed to stem weeks of testing component scenarios saved time.
Improved system reliability – the system relies on data to avoid thermal runaway scenarios in MOSFETs, or inadvertent over-current or spike to low and high voltage systems.
Cost savings in components – based on raw pricing trends, and choices made by other designers, the AI road help favor right high-quality suppliers.
Efficient power systems – sure loss of efficiency reports at the end of the greenThis is why many power applications will drive toward SiC, for instance, EV powertrains, high-performance photovoltaic inverters, and metallurgical process equipment.FAO Section – Common Questions Power Engineers Ask
Q: What things should I be looking out for when selecting SiC components?
A: There are many but focus on thermal performance, voltage rating, switching speed, and desirable reliability metrics (how do they evolve over time).
AI tools will prioritise these and give you ranked options, comparing correlating factors to help cut down the list.
F: How will AI tools assist me in Selection of protection, devices?
A: They will compare fault currents, patterns and environments together with drive type for gates and loads and advise on optimum fuse/circuit breaker type and rating, or advise a hybrid.
F: So I suppose there’s little sense of AI being able to assist me in comparison of suppliers and prices, is there?
A: Wrong! AI will go behind the scenes and look at the suppliers and data on average lead and feedback how happy customers are and suggestion which vendor to use and when.
F: Is AI only useful for large industrial applications?
A: Not at all. While the points made above cater for high voltage/current the smaller print volumes are equally challenging and AI can grown the entire project in its history.
Integrating AI into Your Workflow
To benefit from AI-assisted component selection, power engineers should:
Maintain accurate libraries of components- These should include datasheets provided by manufacturers and historical iterations.
Utilize simulation tools- AI can be used alongside thermal, electrical, and mechanical simulations to aid predicting behavior in new designs.
Iterative optimization – Allow AI to propose alternatives based on changing requirements such as efficiency, costs, operating temperature.
Performance tracking – Ensure information on real-time operation is available to the AI tool so it can refine it’s initial based on real-world use.
Future trends in AI-Powered Power Electronics
Predictive maintenance will become more common, where AI anticipates component failures based on use and history before failover.
Automated multi-vendor optimization will happen, where multi suppliers are being evaluated and compared simultaneously for performance, cost, and availability.
Intelligent thermal and fault management, where the real-time classification of conditions will let it react and adjust to things like overheating or overcurrent conditions.