SUPPORTING companies include:
- Evaluate & validate Machine Learning techniques for Predictive Maintenance on high volume production machinery to deliver optimized system operation.
- Achieve increased uptime & improved energy efficiency utilizing Machine Learning techniques for advanced detection of system anomalies and fault conditions prior to failure.
Today’s methodology of Preventative Maintenance, taking machines offline on a regularly scheduled timeline is not cost efficient and does not necessarily ensure addressing the actual problems leading to system failure. Gaining accurate, actionable insight from the tremendous amount of data acquired in real-time, to understand key component anomalies during operation before system failure, for Predictive Maintenance is a daunting challenge. Furthermore, the root cause of over 80% of failures is not understood.
The discipline of Machine Learning, residing at the edge and up to the cloud, provides a key enabler to Predictive Maintenance through identifying, monitoring and analyzing critical system variables during operation. In the application of Predictive Maintenance using Machine Learning techniques (including supervised and unsupervised) operators can be alerted before system failure, and in some cases without operator interaction addressed, and avoid costly unplanned downtime.
how it works:
This testbed provides the basis for development and evaluation of Machine learning techniques with a focus on the exploration and application of these techniques and algorithmic approaches for time critical Predictive Maintenance and increasing energy efficiency, availability, and lifespan of high volume CNC manufacturing production systems.
The testbed will be developed first in a lab setting, then move into a controlled production environment and finally deployed at an Automotive OEM manufacturing facility.
Predictive Maintenance provides significant value to both companies and people that buy products manufactured by companies employing Predictive Maintenance techniques. Savings achieved during product manufacturing due to more efficient operation and minimizing unplanned downtime could be passed down to customers through price reductions.
- Reduced factory cost to manufacture products and increase competitiveness
- Increased overall operational efficiency with reduction in energy consumption of machine / factory
- Increased understanding of machine manufacturing systems for continuous improvement
- Reduced Risk Priority Numbers (RPNs) of critical components through an increase in detection and reduction in occurrence.
- Random failure detection and resolution through an increase in cause-effect knowledge discovery
- Achieve / Exceed requisite machine line availability (ex. Automotive requires 95% – 98% availability)
- Provide high added value to presently available factory product/process data.
- New services related to MachineLearning techniques and maintenance as a product.
- Machine Learning migration from lab to production facilities.
- Bring competitiveness to production plant based on Machine Learning techniques.
- Higher growth rate potential for companies utilizing Machine Learning-based Predictive Maintenance
Companies are continuously searching for innovative ways to remain competitive by evolving their analytics approach to gain more meaning from the data they are acquiring.
With proper analysis, the information can provide a wealth of insight into company’s system operations, overall operating and maintenance costs.
This knowledge leads to actionable insight, enabling companies to move away from traditional preventative maintenance, with regularly scheduled maintenance times, to that of Predictive Maintenance, for optimized system operation and asset utilization. This testbed focuses on exploring the application of Machine Learning techniques and algorithmic approaches using new innovative technology. This includes Plethora IIoT’s Oberon intelligent systems performing data acquisition, sensor fusion, analytics and Machine Learning powered by Xilinx ZU9 Programmable SoCs for time critical Predictive Maintenance and increasing energy efficiency, availability, and lifespan of manufacturing production systems.
The testbed development is planned in three phases:
- Phase 1: Lab-based development and test facility – Aingura IIoT lab - Spain
- Phase 2: Pilot Factory – Etxe-Tar Manufacturing Facility - Spain
- Phase 3: Automotive OEM Production Manufacturing Facility
Interested in learning more about the Smart Factory Machine Learning Testbed? Email us
- Making Factories Smarter through Machine Learning, IIC Journal of Innovation
- Xilinx Video: Aingura IIoT Analytics Machine Learning
- Webpage: Aingura IIoT
- Technical Paper: Machine Learning Techniques for Industrial Manufacturing