BookmarkSubscribeRSS Feed

Nanyang Polytechnic: CNC Machine Self-Awareness Based on Acoustic Pattern Recognition

Started ‎03-05-2021 by
Modified ‎10-20-2022 by
Views 2,672

 

Machine operators play an important role in monitoring machining performance and identifying behaviour anomalies of CNC machines. Experienced operators can infer the working status and performance based on machining sounds. The nature of machine sounds can be used to determine critical machining events such as collision, tool wear, crash, air-cutting and other behavioral anomalies, etc. With rapid development of IoT, data acquisition capabilities and intelligent acoustic signal processing technologies, there is a growing interest for the development of AI-driven self-awareness system for CNC machines. It will equip CNC machines with ears which can replace the role of machine operators to perform real-time monitoring of the machining job to identify anomalies. It will also play an important role in predictive maintenance and operational efficiency assessment for CNC machines, which is critical to the realization of smart manufacturing in the field of Industry 4.0.

 

In this project, the team will develop an AI-driven machine self-awareness system based on acoustic pattern recognition to automate human intuitive skills to the realization of real-time self-awareness of CNC machines. Critical machining events can then be detected in real-time, i.e. tool crash, tool break and normal cutting. SAS AI edge and toolboxes will be applied to build the pattern recognition neural network for machining events identification based on multi-dimensional acoustic patterns. Supervised machine-learning will be applied to train the pattern recognition neural network with a set of labelled machining pattern data. With the proposed system, CNC machines will be equipped with an intelligent engine with the ability to aware their real-time working conditions. These obtained awareness indicators are helpful in predictive maintenance, operational efficiency assessment and accident prevention of CNC machines.

 

Index Name Role Organization
1 Zhao Zhiqiang Leader Nanyang Polytechnic
2 Zheng Xinhua Data Analyst Nanyang Polytechnic
3 Wayne Wong Chee Weng Application specialist Nanyang Polytechnic
Comments
YWF

Great job, guys!

interesting project team.  question, how did you operationalize the model on the gateway?  Did you use Event Stream Processing?  Thanks.

Fantastic Use case!!!!

But could you please elaborate the model you have built  is really deploy on the live environment.

If the models have deployed then what's the model accuracy in real time.

Are you using any connector or adaptor?

which tools is used  for Visualisation sas VA or ESP Streamviewer or any other tools?

 

Hello uttam631, Thank you for your comments. The model accuracy is still not ideal as we do not have enough training data at moment. We have 500 labelled datasets for 4 machining events while designing the AI model. We have tried different ensemble methods to achieved better predictions. 

Hi, I am the EMEA Marketing Intern based at Heidelberg. We find your story very interesting for the DACH region audience, would you like to do some PR & Marketing with us, if so, please write an email to Angelica.Wong@sas.com 😀

Version history
Last update:
‎10-20-2022 12:32 PM
Updated by:

sas-innovate-2024.png

Available on demand!

Missed SAS Innovate Las Vegas? Watch all the action for free! View the keynotes, general sessions and 22 breakouts on demand.

 

Register now!

Article Tags