How AI, the Edge, and Real-Time Insights Will Fuel Smart Manufacturing

Learn how a Japanese industrial electronics company used real-time insights into its production process to speed up manufacturing and increase throughput. Using artificial intelligence, edge computing and data analytics, Daihen Corporation and FogHorn Systems worked together to eliminate 5,000 hours of manual process and effectively double the Osaka plant’s output.

By Pragati Verma, Contributor

Japanese industrial electronics company Daihen Corporation discovered a problem about three years ago: Every electric transformer manufactured at its Osaka factory required more than 200 manual inspections, and these checkups sucked up 30 percent of the total production time. They realized they needed real-time insights into their production process in order to speed up manufacturing and increase throughput.

Daihen Corporation leaders realized they needed real-time insights into their production process in order to speed up manufacturing and increase throughput.

To get these insights, operations managers installed sensors to monitor the temperature, humidity, and dust levels throughout the manufacturing process via radio frequency identification (RFID). An analytics solution from FogHorn Systems ingested data from sensors and delivered real-time insights into changes that might impact the quality of components being produced. The goal was to produce real-time data and analytics that would track the manufacturing process and identify time spent on each step.

“We knew that harnessing the power of our industrial data would enable us to improve efficiency in factory operations.”

—Ichiro Yamano, executive officer, Daihen

“We knew that harnessing the power of our industrial data would enable us to improve efficiency in factory operations,” says Ichiro Yamano, executive officer for Daihen’s innovation task force. He was right. Within six months of deployment in 2017, the new infrastructure system covered 70 percent of the Daihen’s Osaka factory and eliminated 5,000 hours of manual data entry per year. “We have automated a huge part of the manual process and effectively doubled the throughput of the plant,” says David King, CEO of Foghorn Systems.

Rise of the Edge

The Daihen case study, King says, is an example of how modern industrial Internet of Things (IoT) technology can transform the manufacturing industry. To him, that means big opportunity because industrial assets are an essential part of the world’s gross domestic product and they power the global economy.

YOU MAY ALSO LIKE: How Automotive & Manufacturing Will Change By 2030

One big concern for this industrial sector is lack of technology innovation, he says. Most manufacturing plants, from consumer goods such as automobiles and appliances to industrial machinery and tools, are still running on decades-old machinery. While manufacturers can leverage the technology to reduce labor, material usages and material waste, energy usage, and rework and loss, it isn’t easy to plug their legacy machinery into the internet. “But [advances in ] IoT has now progressed to the point where we can bring the power of artificial intelligence (AI), machine learning , and advanced analytics to the operational technology world (OT). IT is now all set to enter the world of OT.”

“This confluence of 5G networking and emerging technologies such as IoT, edge computing, machine learning, and advanced analytics will turbocharge the move to Industry 4.0.”

—David King, CEO, Foghorn Systems

Analysts at Juniper Research agree and predict 46 billion active industrial IoT connections by 2023. According to King, this growth will be driven by edge computing services that enable these billions of sensors embedded in buildings, production lines, and supply chains to process information and analyze data where it is collected, rather than sending it to a centralized location, such as cloud, for processing. “When we provide the same capability at the edge of the network itself that would normally be done in the cloud, it transforms the entire value proposition,” he says. “It’s cheaper, more secure, and delivers low-latency for on-site data processing and real-time analytics.”

Another factor driving the shift to edge computing is the emergence of new sensing technologies, such as high-resolution videos, imaging, and vibration monitoring. “New technologies produce gigabytes of non-stop data. And you can’t possibly send it all to cloud,” King says. However, he doesn’t see edge computing totally replacing cloud computing. “You will start to have separation of duties… cloud will still be relevant for aggregating and looking across assets.”

The Fourth Manufacturing Revolution

The convergence of physical and digital worlds, also known as Industry 4.0 or Smart Manufacturing, is fueling innovation in several industries.

According to King, solutions like Foghorn will enable manufacturers to roll out new applications, such as advanced monitoring and diagnostics, machine performance optimization, and proactive maintenance. He discusses the first set of applications: condition monitoring, where “you can monitor the condition of assets with a software-defined production control that can be reprogrammed dynamically” and can apply machine learning and deep learning to data sets. “Lots of manufacturing plant technologies exist, but they don’t have the power of complex analytics,” he explains.

Next come self-healing and self-optimized systems, King says, “which use machine learning and deep learning technologies to ingest a non-stop stream of data to make the machines more efficient constantly, as opposed to periodically.” According to King, these self-healing systems can change the manufacturing game. “Industrial systems today perform at 80 to 90 percent efficiency, but they can’t get the last 10 percent because machines break. And the core idea of edge analytics and machine learning is to never let the production system stop. So when the machine self-heals or self-corrects the processes, they will make a tremendous difference in world’s economic efficiency,” he says.

Real-time advanced analytics also enables shop floor managers to optimize asset performance. “Machines often can’t perform at maximum levels, unless they understand the change in conditions in which they are operating,” he says. According to King, the excitement around smart manufacturing is building up. “Several customers are now past the experimental stage and are now going to scale and deploy AI at the edge.”

Spreading the Intelligent Fabric

As this latest industrial revolution moves ahead, fifth generation (5G) wireless network’s high capacity, wireless flexibility, and low-latency performance are promising to accelerate the move to smart manufacturing.” He expects 5G to enable self-healing systems and spread smart manufacturing to remote corners of the world.

“Manufacturing plants are not necessarily in major population centers and connectivity to those sites might not be good today. But 5G now brings all the benefits of machine learning and AI to the whole world,” he assures.

By placing sensors that use high speed and the low latency data transmission of 5G, King expects companies like Daihen to enable AI and automation controls to make quicker decisions. “This confluence of 5G networking and emerging technologies such as IoT, edge computing, machine learning, and advanced analytics will turbocharge the move to Industry 4.0,” he predicts.

“This era of cyber-physical system holds tremendous potential,” King adds. He expects smart manufacturing systems to “make production line faster, efficient, and cost-effective.” And, “as more and more manufacturers invest in digital systems that monitor industrial assets and make decentralized decisions,” they will create new value in an increasingly competitive world.