Daihatsu Uses AI to Classify Engine Sounds
“Although we tried other programming languages, it was hard to implement. We decided to use MATLAB, which allows us to easily import the necessary data by dragging and dropping, and we could easily see the result by ourselves.”
Key Outcomes
- Performed knocking sound analysis with the same accuracy as skilled workers
- Increased AI expertise through MATLAB training
- Promoted visualization of AI and increased awareness of AI
![Daihatsu used AI to identify knocking sounds from its engines.](https://in.mathworks.com/company/user_stories/case-studies/daihatsu-uses-ai-to-classify-engine-sounds/_jcr_content/imageEnhancedParsys/image.adapt.full.medium.jpg/1707462733826.jpg)
Daihatsu used AI to identify knocking sounds from its engines.
In recent years, engines have become more sophisticated, and the number of systems and parts to be analyzed has increased significantly. Daihatsu started to use AI to judge the level of knocking sound, which was previously only done by skilled workers. They needed a tool that could combine deep learning and acoustic analysis.
Also, they urgently need to train AI experts so that data science could be put to practical use.
Daihatsu chose to use MATLAB®, because it has a proven track record in the AI field and it is easy to use, even for those without programming experience.
Using machine learning and feature extraction, Daihatsu easily created classification models, making it possible to examine feature values multiple times in a short period.
As a result, Daihatsu developed AI that can judge the knocking sound with the same accuracy as skilled workers. Working with MathWorks experts, Daihatsu has also conducted data science and AI education with engineers who had not used them before.
Related Resources