The automotive industry is no stranger to big data. For decades automakers have collected and analyzed data from warranty claims, maintenance reports and increasingly smarter, more connected vehicles to improve consumer experience, dealership performance and vehicle quality.
While these activities have been in the public spotlight, big data has also been proliferating on the plant floor as automotive brand owners and tier suppliers embrace digital transformation and smart technology. In the manufacturing space, the challenge is how to best use the massive amount of data smart assets generate to improve short-term and long-term performance.
Sizing It Right
Cloud-based services and analytics platforms have grown up alongside smart devices to help manufacturers capture the value of their digital investments. And some automakers have adopted cloud-based platforms to aggregate, analyze and transform data into powerful business intelligence.
But despite advances, many manufacturers struggle to provide their operators with information that can drive real-time performance gains on the plant floor.
Why? Sending data to a cloud-based platform is ideal for business-level analysis and enterprise-level decision making with more forgiving timelines. But excessive network traffic plus analytic lags can get in the way of delivering timely, contextualized information to the stakeholder who can take corrective action on the plant floor.
In other words, the “analytics to control system loop” is not closed quickly enough to have an immediate impact.
Scalable Analytics: A Faster Way to Optimization at the Edge
A scalable analytics platform closes the loop between big data and the plant floor faster by embedding analytics and machine learning capabilities closest to the source of the information and plant-level decision makers.
For example, a typical automotive assembly plant uses variable speed drives to control motors on material handling conveyors. Modern AC drives continuously monitor output torque and current, which can be directly correlated to the motor’s mechanical parts. The drives can be configured to deliver warnings when parameters exceed limits. In addition, temperature, vibration and other sensors can capture and report critical information regarding gearbox conditions.
Continuous monitoring and analysis of these and other operational parameters can predict gearbox and belt wear or slippage – or motor bearing and winding issues – before causing unplanned downtime. But an optimal maintenance strategy requires timely visibility to that analysis.
A new analytics solution provides one answer at the device level. Delivered on a plug-in appliance, the solution crawls the industrial network and discovers assets – like AC drives and condition sensors. It provides analytics by transforming the data generated into preconfigured health and diagnostic dashboards.
As the appliance uncovers information about how the devices are related to each other, such as fault causality, it starts to understand the system on which it is deployed – and can make prescriptive recommendations. For example, it can send an “action card” to a user’s smartphone or tablet if a drive needs to be reconfigured to maintain optimal performance.
Ultimately, this prescriptive approach enables maintenance teams to be more proactive – and helps minimize potential downtime.
Changing the Game for Automotive Manufacturing
Scalable analytics is a game changer for discrete automotive applications. In addition, this transformational approach promises to be vital in complex continuous processes where machine learning can have a significant impact on product quality and manufacturing velocity.
One example? Prismatic pouch cell battery production. Prismatic pouch cells deliver more energy per volume than their cylindrical counterparts and are gaining traction in the electric vehicle market.
However, prismatic pouch cell production involves a high degree of motion, precision and continuous processing. Optimizing a process in this type of dynamic, multivariable environment is a challenge. But it’s a challenge made for scalable analytics – and machine learning.
Using dynamic mathematical models, the system learns to recognize the impact one variable has on another and automatically adjusts subsequent actions for optimal results. At the same time, the system can deliver critical analytics to operators – such as SPC charts – which enable continual quality monitoring and proactive adjustments.
Keep in mind a scalable approach can extend beyond devices and be applied at the machine and process levels. The platform also can be integrated with MES, OEE and other manufacturing operations and analytics systems to help drive optimization across the enterprise in areas as diverse as production scheduling and energy management.