Rahman Jamal, business & technology fellow at NI, comments.
Markt&Technik: First, a trivial-sounding question: What do you mean by IoT?
Rahman Jamal: IoT is on everyone's lips, but the views are very different, so your question is quite justified. We define IoT as the networking of devices and systems that generate more and more data, so it also includes automated test systems if they are networked.
What importance do you attach to the IoT for industrial production?
We consider the IoT to be the most important technical trend with an impact on industry, alongside 5G mobile radio and autonomous vehicles. We distinguish between two types of IoT: the IoT for commercial applications and the IIoT - Industrial IoT - for smart production, autonomous vehicles and power supply via smart grids, to name just a few examples. In Germany, smart production is mainly understood as automated machines and plants, whereas automated test benches and production testing are not too much in focus. From our point of view, however, they are important, and an Industry 4.0 strategy without automated test benches and without the inclusion of production test data is incomplete. Test asset management will be an integral part of Industry 4.0.
However, the development towards IoT makes design and testing increasingly complex. More and more functions are mapped in software and change rapidly. So the question is: we know what the IoT does with automated testing - but what can the IoT do for automated testing?
A good question - what possibilities do you see?
Automated test benches must be networked in the IIoT just like machines and systems; they generate a lot of data that should be analyzed. We differentiate between different types of data, such as parameterization data, measurement data, application data of the tested product and condition data of the test bench. This data can then be processed with IoT services, which is ultimately the added value of IoT. And depending on the application, you can decide where this should be done: in the IoT cloud platform, in the ERP or MES solution or at the edge. If you make sure that the analytics software is designed for test data, there are many advantages. Test data can also be correlated with manufacturing and design data. In addition, a wide variety of analyses can be carried out - from simple statistics to algorithms used in artificial intelligence or machine learning processes. Common tools such as Python, R and Matlab can also be integrated into the workflow. The prerequisite for this is that the data is networked and can be viewed from different locations. But the test bench itself must also be put to the test and monitored - IoT techniques can be used in this sense.
Meanwhile, commercial IoT platforms are springing up like mushrooms. Instead of inventing everything yourself, you can profit from them in the industry. Their range of tasks is as follows: They grant access to data and take care of so-called "data ingestion", i.e. the prioritization of data sources, the validation of different files and their meaningful forwarding; they manage endpoints and connectivity; they collect, check, prioritize and process data in different formats; they analyse and visualise data; and they develop and manage applications of IoT devices.
What are the prerequisites for automated test systems to do all this?
Currently, most automated test benches are not well networked, although they are more and more globally distributed. They are then isolated solutions. There is also a proliferation of instrument interfaces such as GPIB, LXI and serial protocols. Systems are often distributed in many different ways, and the user lacks an overview of where and in what condition they are. Think of driver and software versions, hardware variants and all these things. In other words, system administration in distributed test systems is a real challenge.
This is where our "SystemLink" comes in, a middleware for the system, data and test management of test benches. SystemLink provides a central interface for the automation of tasks such as software distribution, remote device configuration and system status monitoring. It enables NI and third-party distributed systems to be networked and managed from a single central interface that users can access from anywhere.
What architecture does an automated test system need to be IIoT-enabled?
The foundation for each automated test stand is a software-defined approach with a high degree of modularity. That is the be-all and end-all.
For testing, there is a tendency towards proprietary, organically grown central solutions. A well-structured modular test software architecture with test management, test code, measurement IP, device drivers, hardware abstraction layers etc. would be an excellent basis for exploiting the many advantages of cloud computing.
Based on the different data, the user can apply a wide variety of services for automated testing, ranging from simple messaging services to sophisticated deep learning and AI algorithms that optimize his workflow. It's no coincidence that the future lies in the Internet of Services.
And on this basis, the user can then decide what he wants to move to the cloud, what he wants to pass on to the ERP systems, which reports are to be generated, all the things that are important for business decisions.