The digital health record for transport vehicles and how different symptoms indicate a failure
GlobalFoundries is a global semiconductor manufacturer with 15,000 employees worldwide and more than 200 customers. GlobalFoundries develops process technologies for customers around the world and helps them create the complex and extensive building blocks of technologies that are crucial to how our world lives, works, and evolves.
The Dresden site, Fab 1, is the largest semiconductor plant in Europe, featuring state-of-the-art manufacturing technologies. More than 3,200 employees from around 45 nations work on the company campus. Through its commitment to research and development, GF has helped establish the Free State of Saxony as a leading microelectronics and nanoelectronics hub in Europe.
At the heart of production is the Automated Material Handling System, in particular the overhead transportation vehicles. On a rail network of around 22 km, more than 800 vehicles currently move products every day.
Disruptions in the automation system have a direct impact on core processes and are therefore already monitored by a condition monitoring system. Due to the high mileage of the transport vehicles, mechanical faults occur among other things.
The challenge
The challenge owner’s goal was to implement predictive maintenance for the transport vehicles, enabling failure prediction. Through pattern recognition in the existing alarm system as well as the installation of additional sensors, a health status is to be mapped for each vehicle.
25–35%
The solution from Smart Systems Hub could reduce OHV downtime in the respective problem areas by up to 35%.
Determining a health score based on sensor fusion
Challenge
The challenge (industry challenge) of DPF #4 concerns the driverless transport systems AMHS (OHV) in manufacturing at GlobalFoundries’ Dresden site. These are to be maintained and serviced predictively with the help of an AI-based solution.
Implementation
For this purpose, a Sensor2Cloud IoT infrastructure with a user interface for predicting failures of OHT vehicles was developed. The focus here is on how anomalies in the data from sensors installed in OHVs, as well as in the OHS error logs, can be detected in order to reduce unforeseen stops and downtime through the use of suitable data science or AI algorithms.
Result
With the help of a data collection kit mounted on the rails, acoustic signals from the OHVs are recorded. The classification of the OHVs is based on the acoustic anomalies detected in this way, a “criticality metric” of errors from the existing MACs database, and the detection of similar fault patterns in failed vehicles. Using a clear dashboard, the developed MVP makes it possible to avoid unplanned downtime and realize optimized maintenance planning.
Customer benefits at a glance
Easy to implement at other sites
Applicable to other AMHS components through a modular software and hardware stack
Usable offline (intranet) and online (cloud)
Low energy consumption through vehicle-based recording
Low costs thanks to flexible installation at different sites
Advanced in-cloud data analytics
Many are working on Industry 4.0 solutions, but there is still great potential in Maintenance 4.0. To realize truly predictive and condition-based maintenance strategies, human perception is often lacking. If we digitize the senses, we can stay one step ahead. In combination with AI components, these capabilities can then be made available continuously and flexibly to monitor our system based on its condition. The concept enables us to combine complex capabilities such as human senses and logical thinking and deploy them mechanically. The holistic real-time evaluation during ongoing production will enable us to make statements about the health condition of individual components so that, with efficient “therapy,” the long-term performance of the overall system can be ensured regardless of age and workload.
During Digital Product Factory #4, it was very clear to see what outstanding solutions can emerge when a team of curious experts from different disciplines and levels of experience is brought together by a professional coach and well moderated. Within just three months, a solution was created that works and that we want to implement at our company as soon as possible. Many thanks to everyone involved!
Text taken from the original and translated – Smart Systems Hub
