With the ubiquitous availability of edge devices, robotics, machine learning (ML), and artificial intelligence (AI), manufacturers have tremendous opportunities to redesign—even reimagine—how they do business. Data is at the heart of these opportunities, and it is the driving force for Industry 4.0.
Data drives intelligent operations
The benefits of digital transformation are rooted in the effectiveness of merging AI and ML with data from the factory floor. We see the growing influence of this merger in three key areas of factory operations:
- Quality assurance: Defective products are hard to spot on the production line. Machine vision technology coupled with AI substantially increases the ability to monitor and detect defects on a production line 24/7.
- Predictive maintenance: On-machine sensors and sophisticated analytics provide a constant flow of data that spots possible machine anomalies to address before they become costly repairs.
- Manufacturing productivity: Factories are complex ecosystems of processes, people, and machinery. Intelligent systems that model and simulate process improvements can impact the efficiency and productivity of the entire manufacturing value chain.
These three scenarios are achievable through factory digitization. However, bringing so many assets online introduces the complexity of managing substantial amounts of data. To harness the potential of digitization, manufacturers need solutions that get the right data to the right place at the right time so that systems and people can drive value.
Our new eBook, Turn Siloed Manufacturing Data and Infrastructure into a Competitive Advantage by Infusing it with Intelligence dives into the challenges and benefits of unlocking the potential of this data.
Drowning in data but starved for knowledge
Manufacturing faces multiple challenges with data accessibility. Legacy—often proprietary and nonstandard—devices need to merge with modern technologies. Operations Technology (OT) needs to connect with Information Technology (IT), typically in low-connectivity scenarios.
To address these challenges, technologies are enabling the convergence of OT/IT data. Instead of existing on opposite sides of enterprise firewalls, the technology on the shop floor and the IT department are directly linked. This OT/IT convergence creates a seamless link between devices, processes, and analytics. which provides greater visibility and control. Manufacturers can turn the potential of their data into actionable business intelligence.
Open data standards further support efficient data management. Intel has a long history of driving open standards initiatives across all of our technologies. We started partnering with ExxonMobil in 2014 to define the functional characteristics of an industrial control system. In 2016, both companies helped create the Open Process Automation Forum (OPAF) to develop an open, standards-based process control architecture that allows the industry to rapidly implement innovations, optimize operations, and consolidate existing systems. In our next evolution of open standards, Intel and Red Hat are collaborating to create a new industrial edge platform for building and operating industrial controls.
Moving to the edge for quality control
In manufacturing, data needs to move closer to the source—where reaction times can be measured in milliseconds—to keep up with the speed and continuity of operations needed on the factory floor. For a manufacturing line, this means faster and more accurate insight and the ability to react before anomalies create production losses.
We have seen AI quality control solutions used in everything from heavy equipment manufacturing down to the silicon wafers used in computer chips. They appear in complex autonomous robots and in straightforward manufacturing operations like inspecting the blister packaging commonly used in the pharmaceutical industry. For example, BlueSkies.AI and Intel collaborated on an AI-as-a-service solution to inspect blister packaging moving through a high-speed production line. It is an out-of-the-box solution that doesn’t require AI or IoT skills to deploy. The AI Vision algorithms can be trained to identify defects down to 0.1 mm. The system alerts the line operators to anomalies in real time and the operators immediately identify and remediate the issue. The results are twofold: product waste is reduced, and―most importantly for any manufacturer―production downtime is reduced by spotting machine malfunctions before they cause a shutdown in the line.
Shifting from reactive to predictive maintenance
Asset maintenance has been either scheduled per OEM guidelines or in response to an unexpected failure in the system. Manufacturers are using data to move to a predictive asset maintenance approach to reduce downtime and maintenance costs.
In addition to aggregating data from line assets, gathering environmental data like moisture, air pressure, and temperature or internal asset conditions like engine and oil temperature provides a much clearer picture of how a machine is operating. Because models use data that is current and historical, there is a greater opportunity to fine-tune and refine them over time.
Boosting productivity
Smart factories rely on data to uncover possible productivity gains that range from machine utilization to process workflows to optimized inventory to resilient supply chains. Combining robotics, IoT sensors, machine vision, AI-driven machine learning, and digital twins creates a seemingly endless supply of productivity opportunities.
We see many clothing factories use their data to follow items from raw materials to finished products. Other clothing factories reduce textile defects that are not easily visible to the human eye by 97%. A contact lens manufacturer implemented a solution that enables them to conduct lens inspections 50 times faster with an accuracy rate that is 3 times higher. In industries with low margins, faster and more accurate detection reduces the high costs of defective material.
Intel uses AI to improve productivity across our 10 manufacturing sites. When AI was used to analyze images of wafers to identify defects, we were able to reduce waste, improve production yields, and speed up the production cycle. Intel ships nearly 2 billion units a year across our factories so the bottom-line impacts are incredibly valuable.
Putting data to work
The technology we are developing at Intel around AI, machine vision, machine learning, advanced analytics, and robotics process automation is driving data-driven decision-making, processes, new business models, and ultimately the digital transformation of the manufacturing enterprise.
Read our latest eBook―Turn Siloed Manufacturing Data and Infrastructure into a Competitive Advantage by Infusing it with Intelligence―for a closer look at solutions for unlocking the data needed to create smart factories.
Christine Boles is a Vice President in the Network & Edge Group (NEX) and General Manager of Intel’s Federal and Industrial Solutions. Her organization is responsible for Intel’s NEX Federal and Industrial business within the aerospace, manufacturing, energy, logistics and commercial building segments, including the product and ecosystem strategies for this rapidly evolving space.
Boles joined Intel in 1992 as an application engineer for 16-bit microcontrollers. For almost 30 years, she has led development, delivery and the enabling of customers and ecosystems for Intel based solutions in multiple leadership roles. These solutions span a broad range of embedded and internet of things applications across many industries, including communications, storage, retail, imaging, and commercial buildings.
Boles holds a Bachelor of Science in Electrical Engineering from University of Cincinnati and an MBA from Arizona State University.