While many activities related to the Industrial Internet of Things (IIoT) can seem unfamiliar and unapproachable to less-experienced users, veteran practitioners report it’s mainly an extension of traditional networking and data processing technologies that are relatively easy to practice and apply.
“Clients are still learning and reconciling IIoT with their older factory information systems (FIS), but IIoT is basically the same at heart as we’ve had for 20 years. There’s just more data and more granular details now,” says Dave Fortin, president at DataRealm Inc., a system integrator in Windsor, Ontario, Canada, and certified member of the Control System Integrators Association (www.controlsys.org). “When our company started in 2001-2, we collected information from equipment, and created reports by using desktop PCs as data collector for up to 50 machines. It was connected to a PLC via Kepserver software, which let it pass data to a database on a server, and distribute it to the user’s HMIs via their Intranet.
“This is similar to what we have today, but now an IIoT project has one industrial PC (IPC) for 50-100 assets, and can pass data right up through the architecture. The hardware for IIoT is still similar, but today’s software is very different.”
For example, because DataRealm often works with large automotive OEMs, it’s been collecting data on machine states, conditions and interpretive cycle states for many years. These cycle states typically include bottlenecks, blockages, cycle times and counts, and warnings and faults, as well as documenting traceability for these situations. However, when the system integrator started working on IIoT projects a couple of years ago, it discovered that less software logic conditioning was needed, even though clients wanted more data points for predictive maintenance based on machine learning (ML).
“This type of predictive maintenance takes data over time, categorizes it, and runs it through artificial intelligence (AI) algorithms to predict events, such as the chances of particular failures,” says Fortin. “Better computing and connections by IIoT enable these predictions that allow users to improve operational performance.”
To determine how much computing to do on the plant-floor and how much to do in the cloud, Fortin adds that users have to decide if they’re willing to send data to the cloud. “If the answer is yes, then local or edge computing will be greatly reduced, but data volume and frequency need to be considered,” he says. “True IIoT shouldn’t require data cleaning because the idea is to record, for example, basic machine motions and actions at the lowest level. These can include motor on/off, motor at full speed, part in position, sensor X is on/off, robot is doing Y, or cycle is complete, etc. However, this data does need to be organized to give it context, and this is ideally done on the edge.
“Eventually, all PLC-type input and output data will be collected in near-real-time. This will allow machine control to be emulated, enable any and all states to be created, and let artificial intelligence (AI) analyze entire machine processes.”