Manufacturers can break down intelligence into five different types. The first is business intelligence (BI), historically associated with data housed in business management software. The second is operational intelligence, or the information within distributed systems like a programmatic logic controller (PLC), historians or manufacturing execution system (MES). The third is human intelligence, which refers to the insight from operators that execute the work in the warehouse and manufacturing environment. Combine these three to get a fourth type of intelligence, called collective intelligence, which provides a holistic view of your business. The fifth is artificial intelligence (AI), where we take collective intelligence and apply advanced algorithms to draw correlations and make data actionable to drive continuous improvement and business value.
For some manufacturers, AI can sound too futuristic, scaring them off. Others are interested but don’t know where to start. In fact, 32% of manufacturing professionals ranked not knowing where to start as their first or second highest concern in implementing or exploring digital capabilities like AI.1
Here are some tips on how to leverage AI for manufacturing today:
Begin with a prototype in an area where data is more accessible and well understood. Validate the concept to build momentum and get employees to understand the power of the information. Start with the end in mind by identifying the type of information you need to improve a process and work backwards to understand the data required along the way.
For example, one of the most common data tags that a PLC or control panel has is an emergency stop. By itself, that information doesn’t tell you much. But if you correlate the emergency stop data tag to one or two other data tag values, that information becomes something intelligent and potentially actionable. The emergency stop could be tied to a motor bearing overheating or lost power to a critical component of the unit. Mining into the next level of detail—what triggered the e-stop—gives you time to notify the appropriate people and respond with minimal downtime.
To do this quickly and effectively, tap into an existing in-house or outsourced controls engineering firm or partner to help you navigate the machine output and turn it into operational intelligence.
Take an inventory on the type of investments your company has already made to make data more accessible. You can leverage those investments to incorporate that data in a way that drives actionable insights in the business.
For example, leverage a machine that already has data streaming into its PLC (hint: most PLCs overwrite data every 24 hours). Integrate the most critical data tags into your internet of things (IoT) database. In-Line scales are another source of rich operational data that can help compile valuable metrics regarding specification compliance, “product giveaway” impacting material yields and even production counts.
AI can tap into large and diverse datasets, but sometimes the small things can get overlooked. For example, a label dispenser or carton taper is one of the most fundamental tools on a production line, but they can be used to capture some very critical information. Attaching a simple sensor to this piece of equipment lets you monitor the number of outgoing packages per day, week or month.
Most OEM equipment has its own proprietary communication protocols (the way it transmits and presents data). There are low cost ways to help translate this machine data using common Open Platform Communications (OPC) communication protocols, and those methods can also be used for older analog equipment as well.
Often, an operator or quality technician writes critical information on a piece of paper or enters it into a separate database during the manufacturing process. It can be tempting to focus solely on the data from shop floor machines, but don’t forget that crucial information can come from the human elements of your business processes.
Converting this manual data into a digital format is often critical for continuous improvement, and when using AI algorithms, a manufacturer can correlate this data with automated (machine or device) data to provide powerful insight into operation performance.
To create meaningful collaboration and effective analysis, there must be a certain level of standardization in what people are recording. This allows for information to be aggregated and acted upon quickly.
It’s not uncommon for today’s equipment to have hundreds of different data tags that you can connect and gain access to. Some companies feel the need to connect to and find meaning in every data tag. Although access to all this information seems helpful on the surface, companies often get overwhelmed by the volume and find it difficult to discern which information matters most.
One approach is to apply the Pareto principal and focus on those 20% of data tags that will provide the most meaningful insight and impact 80% of your business operations.
1. NetSuite Manufacturing Survey