Discover, Connect & Collaborate at TECHINNOVATION 2021
Machine Monitoring Using Acoustic Pattern Recognition
An AI based machine self-awareness system for CNC machines was developed. Acoustic data from the CNC machines are detected by IoT sensors and analysed via machine learning algorithms. Sounds associated with various cutting conditions such as normal cutting, heavy cutting, tool wear, and tool crash, can be recognised by the system, and this information can then be feedback to the actuators or end users for necessary actions. The system would facilitate remote monitoring of the machines, predictive maintenance, and failure analysis of the final product. Project could be further enhanced to work for different types of instruments and acoustic sounds.
Technology Features, Specifications and Advantages
The system developed made use of edge AI and software platform for smart manufacturing to enable faster & better decision making. Acoustic data source could be collected conveniently to monitor the real-time machine operation status so that timely actions can be taken when necessary. The system requires minimum interference with the operation and setup of the machines and can be integrated with a wide variety of instruments. The system also offers flexibility to have its AI trained to recognise other types of sounds to address different application needs.
- Precision engineering manufacturing environment that prefers to do predictive maintenance rather than periodic maintenance for cost saving and increase productivity.
- Other applications with acoustic data source for detection & recognition from environment pollution detection to non-intrusively monitoring threatened bird species and water leak detection.
- Equip CNC machines with intelligent engine for real-time awareness and instant notification
- Easy-to-use, economic, and independent to type of CNC controllers and machines
- Solve manpower constraint, improve machine efficiency and factory productivity
- Realize smart machine aligning with Industry 4.0 landscape