The AI Approach to Lean Manufacturing


2020-bl-lean-manufacturing_625wLike many great ideas, lean manufacturing is a seemingly simple idea that proves very difficult to implement. Its shortest definition is manufacturing with the least amount of waste possible. Many more concepts and principles play into that, but what it all boils down to is doing something in the most efficient, least wasteful way. Sounds easy, but when you get into the details of what entails waste and what’s expendable, things can get complicated. That’s where data comes in.

Without data, lean practices or operational excellence efforts are mostly guesswork. Even if a company thinks they’re making a data-based decision, disconnected systems mean the decision is only based on some of the data. And making decisions without all the data is pretty similar to not playing with a full deck.

Lean Manufacturing Requires Big Data

Companies hoping to jump directly onto the lean manufacturing bandwagon need to take a step back for a close, brutally honest look at their existing processes. If a company is still largely dependent on paper or has multiple digital but disconnected systems, data analysis will be more harmful than beneficial. A good example of this is right-first-time (RFT) manufacturing. The idea that you can get it right the first time, every time is really only possible if you’ve correctly identified what holds your processes up and how to correct it. If the data points you in the wrong direction, RFT will remain an elusive goal.

white paper icon

Enjoying this article? You may also enjoy this White Paper:

Bridging the Gap: Manufacturing Software and the Paperless Imperative

Download Free White Paper

Data is essential for process improvement. Artificial intelligence (AI) applications are continually becoming more sophisticated, which means they can offer better insights. The different subsets of AI offer different insights into how to achieve lean manufacturing. Some of these subsets include:

  • Machine learning (ML): using math and statistics to learn from data; improves with experience.
  • Natural language processing (NLP): derives meaning from language.
  • Deep learning: uses artificial neural networks to solve complex challenges.

Machine Learning for Lean Manufacturing

ML in particular shows promise as a means of going lean in manufacturing. One example of this is the ability to avoid nonconformances and deviations. In manufacturing, these holdups require additional paperwork for compliance, and, in some cases, they require corrective and preventive action (CAPA). Reducing quality events goes a long way to reducing waste, both in terms of manufacturing materials and time. ML can also make your manufacturing more efficient by identifying problems in your manufacturing line and suggesting ways to improve.

There is some overlap between ML and NLP and deep learning. They’re not quite like Russian nesting dolls, but they do cover some of the same territory and can work together to produce more insights. For example, customer complaint data is best analyzed using NLP applications. When combined with deep learning, the application can interact with the customers and resolve issues independent of a human. This is most useful for an organization when its quality and manufacturing systems are linked. Customer complaints typically fall in quality’s realm, but they frequently affect manufacturing once the data has been analyzed and the problem identified.

Start Now, but Start Small

We might be talking about algorithms here, but there’s a reason ML uses the word “learning.” It’s surprisingly accurate considering it refers to a program. But learning takes time. We don’t teach kindergartners how to count and expect them to immediately be able to do advanced calculus. The equivalent of teaching an AI application how to count is giving it data. The more data it has, the more accurate the application becomes. Eventually, an application doesn’t just tell you how to be lean by eliminating existing waste, it can predict future sources of waste and tell you how to mitigate them. Instead of just revealing which manufacturing line is least efficient, it can also determine why. When looking at options for how to become more efficient, an AI application removes the guesswork and determines which option is best. The waste accumulated from trying multiple options is eliminated.

Bringing AI into manufacturing is the smart way to achieve RFT and lean manufacturing. However, it all has to begin with getting your data into one location where it can be analyzed. Digitizing your manufacturing processes is a good place to start. With production record data, equipment calibration data, recipe management and more all interconnected, companies can get a holistic view of what’s going on in manufacturing and take steps to create a more efficient, lean manufacturing floor.


2019-bl-author-sarah-bealeSarah Beale is a content marketing specialist at MasterControl in Salt Lake City, where she writes white papers, website landing pages, and is a frequent contributor to the company’s blog, GxP Lifeline. Her areas of expertise include the nutraceuticals, cannabis, and food industries. Beale has been writing about the life sciences and health care for over five years. Prior to joining MasterControl she worked for a nutraceutical company in Salt Lake City and before that she worked for a third-party health care administrator in Chicago. She has a bachelor’s degree in English from Brigham Young University and a master’s degree in business administration from DeVry University.