Leveraging Past Data on the Production Line

June 18, 2019

Industry 4.0 is motivating manufacturing companies plan a bright future where products are made autonomously with minimal human interference, ushering in an era of unrivalled quality and lower costs. A leap this large in efficiency, however, requires large investments and it can be a daunting task to choose where to start. The quickest way to reap rewards could be to look at past data, as most factories already have huge amounts of data flowing in production.

“Great things are done by a series of small things brought together.” — Vincent Van Gough

Even today, many companies already take advantage of data analytics teams to learn from historical production data. This data, however, is usually kept in separate databases which leads to the need to log in to several places to insert or retrieve data. According to McKinsey Global Institute only 20 percent of organisations have a data lake that covers more than half their plants, and only 25 percent use an advanced analytics platform at scale. Such an inefficiency can greatly delay decision making, leading to a loss in quality and time to market.

The first, underlying step of true innovation in production is digitalisation — bringing data to a digital, connected form which a computer can process. This greatly improves data analysis capabilities and helps the team have an accurate, near-real-time overview of the manufacturing process.

Connected historical data provides insight to the cause and effect between the input parameters and their outcomes, influencing environmental effects and differences in the performance of machines. Without being properly connected, such information is almost impossible for engineers to analyse during the production process.

Once a quality dataset is combined and visualised, it’s much easier to pinpoint missing sources of necessary information as well as possible improvements. These improvements can then be automated by applying cutting edge machine learning into the process.

Using machine learning algorithms on top of a quality historical dataset will enable continuous learning. All the data points that otherwise fell on the lap of Engineers and Data Scientists to manually analyse can be automated, freeing up their time to focus on the upkeep and improvements of the production line.

For example, Siemens is using top-down AI to control the highly complex combustion process in gas turbines, where air and gas flow into a chamber, ignite, and burn at temperatures as high as 1,600 degrees Celsius. The incoming data helps them to optimise the work of the turbines in a way that would not be possible without machine learning.

Learning from historical data can seem like a small step to start with but if done right it can be the key in unlocking Industry 4.0 improvements on the production line.

Hedi Hunt is the Senior Product Manager for Production Automation at Neurisium. She helps Customers automate their Production Processes to gain competitive advantage, using her cross industry knowledge of implementing software projects for companies across production, retail and financial industries.

Feel free to also check out our other posts:

Digitalisation: Making Your Data Tell the Truth

Starting with Supply Chain 4.0