Creation of added value for a new service module based on data and algorithms
Information about the machine status in real time
Method for determining structural changes
With predictive maintenance to new customer services: Development of a "health indicator" for Schenck Process Europe GmbH
Schenck Process is a global technology and market leader in the field of applied metrology and scores, among other things, with innovative solutions for weighing, conveying and screening. Schenck Process is planning to implement a use case from the area of predictive maintenance in order to continually develop the range of services it offers its customers. This is particularly relevant for Schenck Process, as unplanned machine failures are associated with high costs and very complex repair processes.
Schenck Process had a limited database of historical data in the run-up to the project. This contained numerous data records of machines in good condition, but only a few data sets with an assignment to specific fault cases. Based on individual specifications, each machine produced is unique.
The goal of Schenck Process is to detect changes in the condition of machines by algorithms. In concrete terms, this involves deriving a “health indicator” that makes the machine status, which is based on sensor data, comprehensible to the user.
In order to create an adequate database, Schenck Process experts defined the current machine states and thus created the basis for training the analysis models. In the context of data management, eoda also created equidistances by interpolating missing measurement periods.
For generating the “health indicator”, eoda relied on methods for the detection of structural changes on the basis of regression models.
Important aspects in the conception of the analysis models by eoda:
- Efficient re-training of the models: Easy transferability of the analyses to other machines.
- Use of real-time data: Status changes in the productive system can be continuously updated.
- Existing toolset: Schenck Process relies on the commercial Matlab software. Therefore, the algorithms and methods used can of course be reimplemented in Matlab.
With the “health indicator”, eoda creates added value for Schenck Process in the form of a new service module. The indicator helps to automatically and proactively detect machine failures and significantly increase understanding of the machine condition.
The “health indicator” can thus make a decisive contribution to increasing machine availability and thus customer satisfaction.