Procter & Gamble’s Jeff Kent on AI and Machine Learning

summary

Jeff Kent, Vice President, Smart Platform Technology & Innovation, Procter & Gamble, spoke at ARC Forum 2023 about his experience applying AI and machine learning. Kent explained the goal of reducing manufacturing operating costs and improving operations, including by 5-10% or more. Improved staffing efficiency, his 50% reduction in maintenance and repair costs, his 50% reduction in QA costs. In addition to 25 years of service as a control and automation subject matter expert at P&G, he also has experience working with the US Air Force on enterprise his systems and networking.

Kent believes AI and machine learning are fundamental elements of the digitalization of production. Mr. Kent has been working on the digitization efforts of Procter & Gamble (P&G) for about five years in a group of companies that started seven years ago. The group has grown to his 25 people and is focused on the practical application of Industry 4.0 concepts, with engineering innovation centers in Cincinnati, Ohio and Kronberg, Germany, near Frankfurt. “It couldn’t be more exciting,” he said.

“Control systems are very powerful assets,” emphasized Kent, noting that P&G aims to deploy thousands of machine learning algorithms at the machine level across 120 sites in more than 40 countries. We have a very important operational intelligence program. “Don’t forget about control,” he said, adding intelligence at this edge level, he explained, is a key part of the IT/OT convergence.

WISE Initiative

The company has created internal WISE branding for this effort, including SmartBox, an edge control and computing device that P&G partners with industry leaders to deliver in a “very practical and affordable way.” Procter & Gamble is committed to deploying AI and machine learning models in all its operations, with a particular focus on the equipment level through coordination across key work systems, such as maintaining quality assurance and utilizing materials. is. Wise is his comprehensive P&G internal service that supports his DevOps and comprehensive machine learning (ML) lifecycle of Smart Box devices.

SmartBox edge device

P&G uses an edge device called “SmartBox” that collects data from existing controllers, new controllers and OEM equipment including interfaces to Mitsubishi, Rockwell and Siemens. Kent notes that many of the algorithms that perform machine learning in P&G’s core work processes need to be tightly coupled in real time with the control system to enable functions such as adaptive control, so the I emphasized that computing is important.

Information is communicated to the OT stack above and below the firewall using OPC UA, and also supports cloud applications in collaboration with Microsoft. Information is made available throughout the organization, accelerating digitization and increasing efficiency, quality and profits. “We are mapping the entire data cycle from data acquisition to data contextualization, model development, model deployment, and information delivery to operators,” explained Kent.

Manufacturing machine learning (ML) lifecycle

The P&G architecture is designed to support the manufacturing machine learning (ML) lifecycle that Kent describes.

• capture data
• Historicalize and contextualize your data
• explore data
• Develop a machine learning model
• Machine learning model testing and validation
• Deployed machine learning model
• Monitor machine learning models
• Maintain machine learning models

OPC UA Foundation

Kent clearly states: “OPC UA is essential. I don’t think there is a better way to establish a common language between OT and IT than with OPC UA.” relationships and seamless data access that scales reliably and repeatably across the organization.

Reach beyond the Purdue model

Reflecting a view shared by many, Kent pointed out the need to change the structural architecture of the system.
“The Purdue model has served us for decades. After all, we have honored it for too long. It’s about starting to blur some of the levels of this to make this a more broadly communicable architecture and an open architecture where everyone can participate.”

“I don’t think we can deliver the power of Industry 4.0 or intelligence unless we introduce a more agnostic network of things. We are at the forefront of what we do.”

Kent further explained: “We are not going to honor the Purdue model. Why should we? Why send information through various layers instead of communicating directly where it is needed? We want to roll it out to all 120 factories.I don’t think there is a better way to establish a common language between OT and IT than with OPC UA.”

---

---