MANUFACTURING, MACHINE LEARNING

Best Uses of ML in Manufacturing from 2021

January 17, 2022

Manufacturing is making a resurgence in the US. The need for a large and cheap labor force drove manufacturing overseas. Now highly efficient and automated factories, with a skilled and tech-savvy workforce are enabling the return of manufacturing to the US. The pandemic and the resulting supply chain crisis will continue to accelerate this move. I see manufacturing leaders continuing to invest in developing/creating a performance-driven and technologically adept workforce.

There is, of course, the concern many have that technology will cost jobs. But not all automation means the use of robots (that replace humans). There are several use cases, where ML is leveraged, to improve product quality and production processes, reduce waste, maintain equipment, increase safety and enhance customer service. To illustrate this, I’m going to highlight a few initiatives we are working on with manufacturing industry leaders. 

  • A global major wanted to label parts coming out of an auto brazer (https://en.wikipedia.org/wiki/Brazing). This is to conduct a traceability analysis and understand factors contributing to defective parts being manufactured. A computer vision solution was deployed to achieve this. Now that same infrastructure can be used to identify a defective part right out of the auto brazer, thereby saving millions in waste. 
  • A 70-year-old manufacturer of specialty ceramic materials wanted to understand how production parameters like temperature, humidity, pressure, etc. affect product quality. The company deployed sensors to capture the relevant data and will use Randomized Control Trials to conduct a causal analysis so that specific parameters can be adjusted to create the highest quality output. 
  • A highly innovative manufacturer of solar tracker systems uses multiple sensors and advanced software to collect a variety of data points from installed trackers. Statistical diagnosis and time series ML modeling will be done to identify patterns causing faults to improve uptime and performance. 

There are other examples of companies using ML to drive innovation in product design, demand forecasting and sourcing, inventory management, and customer relationships. Eventually, these types of solutions will become mainstream and drive a new breed of efficient factories, right here in the US. The resurgent manufacturing industry will continue to drive our economic progress!