How the connected industry can help you predict and prevent failures

Apr. 7, 2017 – History repeats itself, that’s the saying. Well, the same can be said for failure. Despite being 17 years into the 21st century, it is not uncommon to see stories of product recalls, factory failures or other — avoidable — accidents repeated and claiming lives, year after year. Despite technological advances around the world, these incidents remain a regular part of everyday life.

In 2016, this was no different. Samsung was plagued by the faulty Galaxy Note7, resulting in global recalls; airbag manufacturer Takata sparked a global scare after 11 people were killed by defective products in the United States, and Japanese air-carrier All Nippon Airways (ANA) was forced to ground its entire Dreamliner fleet due to defective engines. The list goes on.

But it really doesn’t have to be this way. Technological advances have come far enough to see machines with the capabilities to predict and forewarn where and when things go wrong. Developments in artificial intelligence (AI), machine learning, the Internet of Things (IoT) and data science mean that humans no longer need to rely on their colleagues to spot danger — technology can do it for us. It’s just a matter of adopting these processes and technologies in the first place.

The facts as they stand
This may sound like a pipedream, but it’s not. According to reports from McKinsey, predictive maintenance — the term for using AI and machine learning to prevent future incidents — could save global businesses an incredible $630 billion a year by 2025.

This is because manufacturing machines, and finished products themselves equipped with AI, can learn from previous results by applying cognitive approaches to predictive maintenance. Processes — such as building cars, planes, and even buildings — will reap the benefits of heightened levels of data analysis from machines.

Motor industry analyst IHS Automotive forecasts there will be more than 150 million actively connected cars on our roads by 2020. The combined development of car features and aftermarket devices could result in some two billion connected cars on the world’s roads just five years later. Such a staggering number will result in quantities of data that would make it humanly impossible to detect, isolate and predict incidents.

Conservative estimates predict the average car will produce up to 30 terabytes of data every day. This data is a treasure trove of information regarding the health of the vehicle, including how, when and where the vehicle is driven; the driving style and preferences it is subjected to, and much more.

Only proper analysis can reveal meaningful connections, trends and patterns that can help provide a better driver experience and improve vehicle quality and reliability. This leads to a stronger competitive position and new opportunities for revenue.

How does it work?
What seems like an incredibly complicated process actually is not. Machines with predictive maintenance capabilities monitor results through in-built sensors and interpret this information, with the help of data science, in order to predict when a failure will occur. These warnings can be communicated to human interface monitors (HMIs) long before serious problems manifest themselves. The power of IoT allows machines to communicate quickly and easily with the appropriate people to take the right actions.

The combination of machine learning and AI means that advanced machines can monitor the data from those all important in-built sensors 24 hours a day. It is only by properly employing these vital developments that businesses can reverse their fortunes when it comes to incidents, accidents and recalls.

The possibilities are limitless
And it isn’t just about machines on factory floors giving advance warnings about potential danger. Predictive maintenance can also help minimize downtime on the plant floor and increase productivity across the board, saving time and money for all.

The fact is that by combining telemetric data and predictive maintenance, manufacturers can maximize the lifetimes and usefulness of the machines they have at their disposal. By combining data provided by sensors — on products, machines or vehicles — with predictive techniques, this data can be used to capture the real-time status of parts and functions. This information, more easily harvested by adopting data science, machine learning, AI and IoT, can be used to improve safety and give human overseers a “heads up,” solving future problems before they even occur.

Sundeep Sanghavi is the co-founder and CEO of DataRPM, an industry pioneer in cognitive data science. He’s a highly accomplished entrepreneur and data junkie with more than 20 years of experience using data as the currency to perform advanced analytics.

This feature was originally published in the March/April 2017 issue of Manufacturing AUTOMATION.