Recently, our colleague Erik Holleboom travelled to Budapest to hear and learn about Machine Learning. The reason for this was to expand the expertise of Strypes in this field. Our colleagues build controllers for mission-critical systems and those systems require high availability. Availability is promoted when maintenance/management is carried out on time or even at the most ideal time. But how do you predict when the ideal moment is? And how do you make this applicable? In other words: how do you realise knowledge for predictive maintenance?

Smart controls

The controls we build generate a lot of data. This data is of course stored and registered, but we do not know whether this is the right data. If the data is analysed intelligently, this can provide the information required to make the prediction for maintenance and management. We also learn which data is relevant and which is not. Analysing this data manually takes a lot of time. An AI system offers the possibility to do this more efficiently and on a larger scale. The algorithms required for this already exist. However, the challenge here is to creatively apply those algorithms, put together the right training sets and let the system learn. Once the system has ‘learnt’, it can be used to actively predict, also for example, detecting anomalies.

In practice

What does that look like in practice? An example: the controls we build for flood defences provide information about water levels. With this and other information together, we can better predict future water levels, for example, which means that doors will automatically close when high water levels are expected. Up until now, it was the operator who received or retrieved information to determine whether the barrier doors should close. With the Machine Learning application, the system itself will ‘know’ when to act. This means that the doors close automatically when the water level rises due to the tide and the doors remain open when a wide ship causes temporary higher water levels. The system will therefore, using human input, ‘learn’ how to interpret the data, enabling it to differentiate the situation, recognise false positives or true negatives and take appropriate action.


Missing link

When looking at Artificial Intelligence, data collection, IoT devices etc., one often starts with the technological aspect. In other words, devices are linked to the system to collect data. This generates, together with the data already present in the operating systems, an enormous amount of data. This data then has to be analysed and the meaningful data is visualised. At the moment, three steps have already been taken to achieve a complete smart system, but no profit has yet been made.

From our experience we often see that such projects start to work in this way full of enthusiasm. After a while, it turns out that an enormous investment has been made without a clear view of the result. We therefore believe that, together with our clients, we must first look at the business case. Together we go through the entire process top-down and then implement it bottom-up. This makes it clear to everyone which investment is needed and when profits can be made.

Strypes makes a difference

Within our vision, an organisation that really makes a difference is aimed at serving customers as well as possible and ensuring that employees develop to the maximum and feel involved in the company. Growth and profit for the organisation are derived from growth and profit for our clients and our employees, not a goal in itself. Wondering how Strypes can help your organisation improve? Get in touch.

CONTACT BY E-MAIL +31 (0)343 769024