Next-generation smart factories will use what’s known as a cyber-physical system to collect and analyze data to carry out tasks more efficiently and create better products.
Our phones, cars and water have all gotten smart, so it was only a matter of time before industry did, too. You’ve probably heard the term smart factory, or maybe you’ve heard about the industrial internet of things (IIoT). It’s simple to deduce that these next-generation factories introduce new technologies to do things even better. So, how exactly do they do it?
A combination of interconnectivity and tools known as a cyber-physical system (CPS) is what makes a smart factory “smart”. They allow tasks to be carried out more efficiently by collecting and analyzing data. That information can then be used to create better products and more efficient techniques, sometimes by the factory itself.
Hungry for Data
The first step in making a factory smart is to centralize your data. Any successful business should be run with a keen eye on the numbers. In a smart factory, the proper systems are in place to collect and centralize those numbers. In most cases, this is a network of wireless IIoT sensors and devices that are constantly collecting and storing vast amounts of data. This data could be anything from the timing of specific robots to environmental conditions throughout the factory.
Without a good set of aggregated data and the right tools to maintain it, you cannot have a smart factory. Having the data, however, is just the most basic component of a smart factory.
In fact, the smart factory model categorizes available data as level one of four. In this level, data is being captured but in a way that makes it difficult to analyze holistically. Individual processes or systems hold on to the data they collect, and you can’t really get a birds-eye view of factory processes — even if you go to the trouble of collating all the data that’s available.
Level two is accessible data, which means that the data has been pulled out of the silos or disparate corners of the company. The data goes into a central reporting structure of some kind, which employees can then use to make informed decisions. At this point, the kind of data being collected may be standardized so that it’s easier to analyze.
Making Informed Decisions
Having lots of data isn’t enough, though. You need to know what to do with it. This is where the smart factory model begins to shine. As more data is aggregated, it becomes possible to create models of processes and subprocesses that inform the factories’ overall missions. Network-aware CPSes communicate what they’re doing and the outcomes of their tasks via the Internet of Things (IoT). As a result, the factory itself becomes aware of its own successes and failures, which are defined by people in administrative roles using key performance indicators (KPIs).
In level three of the four-level smart factory model, the smart factory manager has implemented advanced data analytics technology powered by artificial intelligence (AI) and big data analysis. These technologies can sift through huge amounts of data — more than even a team of human statisticians could reasonably analyze — and detect patterns and create predictive models.
Factory managers and line employees can then use these insights to improve factory processes or make more informed decisions.
At the highest level of a smart factory — level four — the factory’s use of data itself has become automated. Factory processes can effectively recognize when something has gone wrong and propose new ideas to generate better outcomes — sometimes, without any human intervention.
Machine Learning and Execution
The operation is self-aware enough to halt production or make a change when something goes wrong. For example, in automated parts inspection, a process used by major manufacturers around the world, defective parts are thrown out immediately rather than submitted to paid human QA testers. It saves considerable time and money.
The parts inspection works by using a model that tells the CPSes exactly how a finished part should look, measure or perform. The robotic CPSes can then test the piece and make a decision on the fly while recording examination outcomes, which are cataloged — and might even be added to a data set of “bad” examples, improving the parts inspection algorithm.
Let’s use the example of an automotive factory to further illustrate how a level four smart factory works. Imagine an operation that’s configured to maximize the number of cars produced every day. The CPSes in the factory require maintenance every 100 vehicles, which requires you to take some or all of the factory offline to carry out. The smart factory has a feature that notifies the administrator when maintenance is required or even begins the maintenance process itself.
This operation uses a lot of power, but it can allow factory managers to save on the cost of electricity by leveraging the capabilities of their smart factory. The factory is context-aware, which means that, by defining power usage costs as a KPI, you can configure it to use electricity as efficiently as possible. You can conduct maintenance later in the day during off-peak hours, or slow the rate of production to do less work when power is expensive.
Because of the interconnected nature of smart factory systems, the factory can make recommendations to managers and even take action on its own.
Does this mean factory workers can kiss their jobs goodbye? It’s more likely that jobs will be created in other parts of the operation, so there may be a shift in where efforts are placed. That means there will likely not be a decrease in the number of positions.
Do you work for a business that could benefit from this type of technology? How do you think smart factories will change the face of industry in the future? Tell us in the comments section below.