News, Product, Event

Mathematical models to improve predictive maintenance forecasting

Björn Lorenz08/17/2018

Predictive maintenance is one of the most important areas of application in mechanical and plant engineering for Internet of Things (IoT) technologies. That was reason enough for the FASA e.V. to put the topic on the agenda of the 29th  meeting of the industry working group on cooperation in plant engineering" which met last week in Magdeburg. The FASA e.V. is an association for the promotion of mechanical and plant engineering, which is supported by the Fraunhofer IFF research institute and COSMO CONSULT TIC GmbH, the think tank of the COSMO CONSULT group. 

The IoT requires powerful data analysis

Predictive maintenance entails continuously monitoring critical parts subject to wear using sensors in order to prevent cases of unplanned maintenance. Along with the sensors, the software plays a key role, because the sensors provide an enormous amount of data, of which only a small portion is usually needed. Data analysis solutions help filter out the relevant values and then present them in a way that is useful for making decisions. "We are only interested in certain events - for example, when critical limits will be exceeded that indicate a failure will occur," explains Stefan Rohkohl, managing director of COSMO CONSULT TIC GmbH, adding "this information should be available quickly and presented flexibly to the necessary depth, for example mobile as graphical analysis on smartphones or tablets, which is one of the key demands of the working group."

Mathematical models to improve forecasting

COSMO CONSULT Data Science went a step further. The company, which specializes in mathematical optimization processes, presented mathematical models that can be used to take better advantage of the sensor data: Whereas in the course of data tracking only real events - such as the exceeding of a previously defined limit value - trigger action, you can also use mathematical models to calculate probabilities of occurrence for particular situations using patterns. In this way, it is possible to forecast imminent failures or determine optimal maintenance cycles - without something having to happen first. Machine downtime and deviations in quality in production can be significantly reduced. In addition to this, due to better planning capabilities, maintenance measures can be bundled, which reduces the costs for parts subject to wear and for service technicians without affecting the production process. "Our forecast models cover the life cycle of vehicle or machine parks. We aggregate data, create transparency and identify dependencies at an early stage. The mathematical optimization models we have developed - such as for data mining and operations research - are tightly integrated. They pave the way for data-driven processes and provide demonstrable cost savings immediately after being implemented," says Roland Abele, managing director of COSMO CONSULT Data Science GmbH.

Other presentations at the 29th meeting of the industry working group, which ended late Tuesday afternoon, were on topics such as electronic batch tracking for flange gaskets, network infrastructure in industrial parks, and innovative concepts for machine and plant safety testing.

Are you also thinking about how you can maintain your machines and systems more efficiently with the help of predictive maintenance? Then let us talk about it. Just call us or send us an e-mail.

Share post

Tags

No tags available.