Is AI the new brains of manufacturing?

From cars to mobile phones, more and more devices are able to communicate with us and with each other. The same applies for manufacturing plants and production machines where huge amounts of data are available. Is advanced artificial intelligence (AI) the solution that can make sense of all the data from the shopfloor, so that it becomes useful for optimization and decision-making? And will AI be the next step towards the digital transformation of industry?

Artificial intelligence (AI) is a collective term that encompasses various technologies, including machine learning, neural networks and natural language processing. It differs from conventional computing in that systems no longer need to be programmed every step of the way – they learn for themselves. Broadly speaking, the aim is to create machines that can sense their environments, and make judgements and decisions on their own which extend todays known intelligence of machines and brings more freedom to production.

What was once unknowable becomes known

Modern AI excels at extracting useful information from vast amounts of raw data, and this capability means the technology will play a key role in many areas. The result will be that all sorts of knowledge that was previously vague, or non-existent, will become increasingly quantifiable and precise.

I believe that in factory automation, AI will be most valuable through machine learning algorithms that process the volumes of product and process data, and facilitate continuous learning and improvement. It will significantly reduce the programming and engineering costs. And over time, as the machines gain experience, they can continuously improve their own abilities, with little or no need for human intervention.

AI has proven its value in certain industrial applications already but not in scale. The first proper industrial applications generally involved performing tasks such as documenting surroundings, tracking inventory and analyzing factory data to improve energy efficiency. In the future, AI will be capable of much more, such as utilizing a digital twin of a manufacturing process to determine whether the product that’s being manufactured meets quality standards – and if not, adjust production parameters automatically to resolve the issue.

In this regard, AI can make production more reliable, more efficient, and companies will be more competitive. Or, to put it another way, AI has the potential to lift competitive standards to such a degree that no one can afford to do without it.

How AI helps industry

AI is expected to be involved in everything, from supporting design teams to empowering control systems. However, while the potential is clear, actually applying AI to manufacturing operations is not so easy. Companies have individual needs, and most will need a partner with the expertise and resources to train the algorithms effectively. Siemens aims to provide its customers with automation solutions that make accessing AI technology as simple as possible.

A case in point is Edge computing. As I discussed in the previous blog, Edge is an enabler for AI, and for many manufacturers, it is the most promising path into the digital transformation, especially when Edge is combined with the cloud. The Edge platform makes data available more quickly, while the heavy-duty computing power comes from the cloud solution.

Siemens is leading by example at its factory in Amberg, where millions of SIMATIC components are manufactured each year. In the past, every printed circuit board (PCB) had to be x-rayed to ensure 100% quality before being shipped to the customer, and this time-consuming process naturally limited plant productivity. Now, AI makes it possible to assess the quality of each PCB during the manufacturing process and predict where faults, if there are any, will turn up.

An algorithm has been trained to interpret the quality-relevant process data of the PCBs. And with the help of a model that runs on an integrated Edge application, the algorithm predicts which PCBs are fault-free and which ones aren’t – in other words, whether an x-ray test is necessary or not. This AI application has boosted efficiency by around 30 percent, and as the algorithm learns, its predictions get more and more precise. Learn more.

Security comes first

In order to take part in Industry 4.0 and the digital transformation, every company – regardless of size – needs a state-of-the-art, end-to-end IT infrastructure. But, of course, no company will integrate a solution into their environment if they don’t have complete confidence in its security.

On the one hand, having more and more intelligent devices and integrated networks greatly increases the number of potential targets that need to be protected. AI, however, can also enhance cyber security, particularly in terms of risk detection, by doing the same things it does for production – by analyzing large amounts of data and detecting anomalies in real time. 

Security is another aspect in which Edge may provide a way in for many companies. Most large corporations have reportedly begun utilizing AI technology in one form or another, but for others – particularly SMEs – it may still seem very abstract and a bit daunting. Enabling AI to learn requires a great deal of computing power, which is generally only available in the cloud. Using an Edge platform, which is integrated into the customer’s own environment, means a connection to the cloud is only necessary for specific tasks. This offers the benefit of computing power while allaying potential security concerns about connecting with cloud solutions.

AI has been the subject of decades of work and has come a long way. What sounded like science fiction a couple of decades ago is becoming a tangible reality today. Although AI can already outperform human brains at a growing range of tasks, it is unlikely that the machines are going to lock us out of the factory any time soon. Ideally, AI will enhance traditional work in a number of ways, such as relieving us of many repetitive tasks, providing early warning of potential issues, and enabling much faster response times and better-informed decisions.