Complementary learning for AI-based predictive quality and maintenance

Using machine learning, deep learning, and cognitive computing in concert can help enterprises gain competitive edges.

By Alice LaPlante
May 3, 2018
Fractal kaleidoscope Fractal kaleidoscope (source: Pixabay)

The cost of unplanned downtime due to equipment failure across all industries is huge. Back in 2014, Aberdeen estimated the cost to businesses to be $164,000 an hour, on average. By 2016, that figure had skyrocketed by 60% to $260,000 an hour. Monetary losses aside, the impacts of unplanned outages for a business can be devastating—according to a 2017 study by Vanson-Bourne, 46% of businesses said they were not able to deliver services to customers and 37% said they lost production time on a critical piece of equipment or other asset.

Take the airline industry as an example. Flight delays due to mechanical problems are common. Sometimes the issue gets fixed in timely manner. But sometimes it takes hours, and, in worst-case scenarios, flights are canceled. The costs to the passengers can be high—missed connections, missed meetings, missed vacations, and more. The cost to the airline is immense: lost revenue, lost customer satisfaction, and lost brand value in addition to all the money—labor and parts—it takes to fix the problem. Consider the overtime, the shipping costs of rushed parts, and all the lost revenue, and you see how it adds up. In fact, unplanned maintenance of equipment costs three to nine times more than planned maintenance.

Learn faster. Dig deeper. See farther.

Join the O'Reilly online learning platform. Get a free trial today and find answers on the fly, or master something new and useful.

Learn more

Clearly, a lot is at stake. Which is why many industries that focus on product quality and maintenance are turning to next-generation artificial intelligence (AI)-based predictive quality and maintenance (PQM) solutions.

In this article, I’ll go over what PQM solutions are and explain how they are being made dramatically more effective by leveraging AI. I’ll also explore complementary learning and look at how today’s leading AI-based PQM solutions incorporate it. Finally, I’ll give a real-world example of AI-based PQM in action.

What are PQM solutions?

PQM solutions, which harness data gathered by both the Internet of Things (IoT) and data from traditional legacy systems, focus on detecting and addressing quality and maintenance issues before they turn into serious problems—for example, problems that can cause unplanned downtime.

Unplanned downtime is a major cost driver in any industry that must maintain large inventories of capital assets. For an airline, for example, delaying flights due to unplanned maintenance can cost thousands of dollars each minute. Unplanned shutdowns of oil platforms can run into the millions of dollars. And in manufacturing plants, the costs of disruptions go directly to the bottom line. It is the goal of every organization to eliminate unplanned downtime in favor of planned maintenance. PQM solutions can help with planned maintenance, too, by shortening maintenance operations windows.

PQM solutions help businesses prioritize how to best allocate scarce resources, resolve issues, and plan ahead, thereby keeping—or achieving—a competitive edge. Recent research suggests that the market for PQM solutions will grow from $2.2 billion in 2017 to $10.9 billion by 2022, a 39% annual growth rate. And, of the top 10 uses predicted for AI in 2025, PQM comes in fourth place.

Artificial intelligence enters the picture

Traditionally, PQM solutions crunched numbers and came up with average statistics to predict when quality corrections or maintenance were required.

Now, with the availability of much bigger data sets and the latest developments in AI, it’s a whole new game. AI-based PQM solutions differ from traditional predictive quality and maintenance solutions because they analyze the actual condition of a product rather than just using average or expected statistics to predict when quality corrections or maintenance might be required.

AI-based PQM solutions use several technologies in concert, including machine learning, deep learning, and cognitive computing:

  • Machine learning: focuses on real-world problems by processing—and learning from—large amounts of data
  • Deep learning: uses neural networks to be able to sort through nearly unimaginable volumes of data to come to conclusions
  • Cognitive computing: a subset of AI that attempts to mimic the way humans think. And a very important subset of cognitive computing is associative-memory learning and reasoning, which mimics the way humans learn, remember, and reason by making associations

Complementary learning: The future of PQM solutions

Because each type of AI is good at solving different problems, applying them simultaneously is the key to success. Complementary learning in the context of PQM applications involves combining all these types of AI—machine learning, deep learning, and cognitive computing—to get insight into quality and maintenance issues.

In effect, a PQM solution that embraces complementary learning first uses machine learning and deep learning to answer the question, “What is the problem?” Then, cognitive computing answers such questions as: “Have I ever seen this before? What type of a problem is it? Who knows how to fix this? What caused this problem? And will it happen again?”

AI-based PQM solution in action

Accenture is a global professional services company that provides a broad range of services and solutions in business strategy, consulting, digital, technology, and operations. One of its many services focuses on software testing.

This is a rich area for AI-based PQM solutions for a couple of reasons. First, many organizations have not changed their software-testing techniques in the past 20 years. Thus, software testing often involves a significant amount of manual labor. For some mission-critical systems, test engineers spend up to 90% of their time managing test cases and documenting them rather than actually testing.

Second, the overall quality of software has traditionally been treated as just one isolated component in the software development life cycle. For example, QA used to be considered a step outside of the product development cycle—involve QA when everything is done. But testing can no longer be performed in silos outside of the development process. Now, with devices and connections getting more complicated, QA needs to be integrated as part of the software’s design and production process. If QA sees a trend that a certain part or certain function causes frequent issues, they can report such a trend to the product team to find the root cause and fix.

Accenture’s platform uses Intel Saffron AI in its software testing solution to accelerate automation, spot trends, manage risks, and continuously respond to customer feedback. According to Accenture, this approach has reduced time-to-market and cost of testing by more than 20%, and can also save 30% to 50% of time spent over-engineering.

“Testing is transforming into quality engineering, where artificial intelligence-driven analytics is at the core of driving productivity and agility,” says Kishore Durg, senior managing director and global testing lead for Accenture. “That is what the Accenture Touchless Testing Platform is designed to do.”

Looking ahead

AI, specifically in the context of complementary learning, will play a huge role in future PQM solutions that will make up for the fact that human resources simply cannot scale at the same rate as data. AI promises to extend people’s capabilities, and with a growing number of out-of-the-box vertical solutions becoming available, it’s easy to see why AI-based PQM solutions are projected to grow at such a rapid pace in the not-too-distant future.

This post is a collaboration between O’Reilly and Intel Saffron. See our statement of editorial independence.

Post topics: Artificial Intelligence