AVL Focus - Issue 2024

ecent advancements in artificial intelligence algoritLms

and Lardware Lave significantly imTroved simYlation

and testing processes. Within this context, three AI

applications stand out:

ƍ %- EnEP]tiGW Leveraging decision trees and neural

networOs, %- simTlifies Tredictive maintenance on testFeds

and the creation of Digital Twins.

• +enerEtive %- Inspired by ChatGPT, generative AI enables

easy chatbot creation to provide engineers with expert tools

for their daily work. It also extracts relevant information

from the wealth of data generated on testbeds, aiding in

UYicO identification oJ areas to JocYs on.

ƍ 'oQTYter viWion By analyzing the integrity of the unit

under test (e.g., monitoring battery thermal runaway),

computer vision enhances safety and reliability.

As the number of algorithms and tools proliferates, thoughtful

analysis of requirements and available hardware becomes

essential for achieving cost-effective performance. For

predictive maintenance on testbeds, AVL has opted for

491%ɸ 1acLine 0earning witL decision trees dYe to tLeir

suitability within hardware and performance requirements.

Additionally, to optimize resource consumption and obtain

understandable results, raw measurements from the

testbed undergo real-time preprocessing using advanced

matLematical oTerations sYcL as sTecific filters and **8 .

AI algorithms then analyze this preprocessed data. The

testbed performs a live comparison of raw measurements

and predictions, and compares predicted vs. calculated KPIs

Larmonics, eƾciency, state oJ LealtL, etc. .

Furthermore, a transfer learning approach using the Bayesian

model helps translate KPIs into recommendations that

testbed operators understand without involving experts.

Integrating modern AI algorithms into testbeds, combined

with mathematical and statistical analysis, aids in identifying

problems and quickly understanding the root cause of issues

with the unit under test.

The automotive and aerospace industries are actively pursuing cost

optimization and propulsion system efficiency. This drive necessitates

the evolution of electric motors – emphasizing compactness, higher

speed, and power density. Components such as bearings, coils, and

magnets experience significant stress, underscoring the critical

importance of meticulous design.

How to Accelerate

E-Motor Development

and Testing with AI

2024