Connecting Data from Sensors to AIoT Systems with ease

11 September 2019 | Web Article Number: ME201916174

Instrumentation, Measurement & Control

KPMG, a distinguished global firm providing audit, tax, and advisory services, leveraged AIoT (AI and IoT) technology to help an automotive engine parts manufacturer to increase yield and build predictive maintenance.

More sensors were added to existing IoT devices to collect additional data on vibration, temperature, rotating speed, and electric current. Moxa's easy to-use connectivity solutions were implemented to help send the data to a backend AI platform where, through analysis, control standards were established, making predictive maintenance possible as any deviation, which could result in the production of defective products, was immediately detected. The OEE was increased from 70% to 85%.

Predictive maintenance enables operators to predict when maintenance should be performed by determining the condition of their in-service equipment. Basically, predictive maintenance comes down to keeping equipment in good working order to prevent unexpected downtime, thus ensuring reliability.

This practice brings huge cost savings in comparison with routine or scheduled preventive maintenance, because tasks are performed only when needed. ARC Advisory Group estimates that predictive maintenance can reduce maintenance costs by 50% and unexpected failures by 55%.

To further increase the efficiency of predictive maintenance, it is critical to leverage the ability of edge computers to pre-process increasing volumes of data acquired from sensors, meters, and other network devices, as well as to autonomously react before machine failures occur. However, with new opportunities come new challenges, and the same holds true for enabling predictive maintenance.

Two main challenges that managers have to deal with are performing diverse data acquisition and deploying edge intelligence.

Paper presented by Robert Wright of RJ Connect at a recent Durban SAIMC meeting

Related Articles